Pub Date : 2024-05-15DOI: 10.59275/j.melba.2024-7189
Ivan Diaz, Mario Geiger, Richard Iain McKinley
Convolutional neural networks (CNNs) allow for parameter sharing and translational equivariance by using convolutional kernels in their linear layers. By restricting these kernels to be SO(3)-steerable, CNNs can further improve parameter sharing. These rotationally-equivariant convolutional layers have several advantages over standard convolutional layers, including increased robustness to unseen poses, smaller network size, and improved sample efficiency. Despite this, most segmentation networks used in medical image analysis continue to rely on standard convolutional kernels. In this paper, we present a new family of segmentation networks that use equivariant voxel convolutions based on spherical harmonics. These networks are robust to data poses not seen during training, and do not require rotation-based data augmentation during training. In addition, we demonstrate improved segmentation performance in MRI brain tumor and healthy brain structure segmentation tasks, with enhanced robustness to reduced amounts of training data and improved parameter efficiency. Code to reproduce our results, and to implement the equivariant segmentation networks for other tasks is available at http://github.com/SCAN-NRAD/e3nn_Unet.
{"title":"Leveraging SO(3)-steerable convolutions for pose-robust semantic segmentation in 3D medical data.","authors":"Ivan Diaz, Mario Geiger, Richard Iain McKinley","doi":"10.59275/j.melba.2024-7189","DOIUrl":"10.59275/j.melba.2024-7189","url":null,"abstract":"<p><p>Convolutional neural networks (CNNs) allow for parameter sharing and translational equivariance by using convolutional kernels in their linear layers. By restricting these kernels to be SO(3)-steerable, CNNs can further improve parameter sharing. These rotationally-equivariant convolutional layers have several advantages over standard convolutional layers, including increased robustness to unseen poses, smaller network size, and improved sample efficiency. Despite this, most segmentation networks used in medical image analysis continue to rely on standard convolutional kernels. In this paper, we present a new family of segmentation networks that use equivariant voxel convolutions based on spherical harmonics. These networks are robust to data poses not seen during training, and do not require rotation-based data augmentation during training. In addition, we demonstrate improved segmentation performance in MRI brain tumor and healthy brain structure segmentation tasks, with enhanced robustness to reduced amounts of training data and improved parameter efficiency. Code to reproduce our results, and to implement the equivariant segmentation networks for other tasks is available at http://github.com/SCAN-NRAD/e3nn_Unet.</p>","PeriodicalId":75083,"journal":{"name":"The journal of machine learning for biomedical imaging","volume":"2 May 2024","pages":"834-855"},"PeriodicalIF":0.0,"publicationDate":"2024-05-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7617181/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142803819","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-12-01DOI: 10.59275/j.melba.2023-g3f8
S Mazdak Abulnaga, Neel Dey, Sean I Young, Eileen Pan, Katherine I Hobgood, Clinton J Wang, P Ellen Grant, Esra Abaci Turk, Polina Golland
Blood oxygen level dependent (BOLD) MRI time series with maternal hyperoxia can assess placental oxygenation and function. Measuring precise BOLD changes in the placenta requires accurate temporal placental segmentation and is confounded by fetal and maternal motion, contractions, and hyperoxia-induced intensity changes. Current BOLD placenta segmentation methods warp a manually annotated subject-specific template to the entire time series. However, as the placenta is a thin, elongated, and highly non-rigid organ subject to large deformations and obfuscated edges, existing work cannot accurately segment the placental shape, especially near boundaries. In this work, we propose a machine learning segmentation framework for placental BOLD MRI and apply it to segmenting each volume in a time series. We use a placental-boundary weighted loss formulation and perform a comprehensive evaluation across several popular segmentation objectives. Our model is trained and tested on a cohort of 91 subjects containing healthy fetuses, fetuses with fetal growth restriction, and mothers with high BMI. Biomedically, our model performs reliably in segmenting volumes in both normoxic and hyperoxic points in the BOLD time series. We further find that boundary-weighting increases placental segmentation performance by 8.3% and 6.0% Dice coefficient for the cross-entropy and signed distance transform objectives, respectively.
{"title":"Shape-aware Segmentation of the Placenta in BOLD Fetal MRI Time Series.","authors":"S Mazdak Abulnaga, Neel Dey, Sean I Young, Eileen Pan, Katherine I Hobgood, Clinton J Wang, P Ellen Grant, Esra Abaci Turk, Polina Golland","doi":"10.59275/j.melba.2023-g3f8","DOIUrl":"10.59275/j.melba.2023-g3f8","url":null,"abstract":"<p><p>Blood oxygen level dependent (BOLD) MRI time series with maternal hyperoxia can assess placental oxygenation and function. Measuring precise BOLD changes in the placenta requires accurate temporal placental segmentation and is confounded by fetal and maternal motion, contractions, and hyperoxia-induced intensity changes. Current BOLD placenta segmentation methods warp a manually annotated subject-specific template to the entire time series. However, as the placenta is a thin, elongated, and highly non-rigid organ subject to large deformations and obfuscated edges, existing work cannot accurately segment the placental shape, especially near boundaries. In this work, we propose a machine learning segmentation framework for placental BOLD MRI and apply it to segmenting each volume in a time series. We use a placental-boundary weighted loss formulation and perform a comprehensive evaluation across several popular segmentation objectives. Our model is trained and tested on a cohort of 91 subjects containing healthy fetuses, fetuses with fetal growth restriction, and mothers with high BMI. Biomedically, our model performs reliably in segmenting volumes in both normoxic and hyperoxic points in the BOLD time series. We further find that boundary-weighting increases placental segmentation performance by 8.3% and 6.0% Dice coefficient for the cross-entropy and signed distance transform objectives, respectively.</p>","PeriodicalId":75083,"journal":{"name":"The journal of machine learning for biomedical imaging","volume":"2 PIPPI 2022","pages":"527-546"},"PeriodicalIF":0.0,"publicationDate":"2023-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11514310/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142523827","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-11-11DOI: 10.59275/j.melba.2023-b7bc
Paula Ramirez, Alena Uus, Milou P.M. van Poppel, Irina Grigorescu, Johannes K. Steinweg, David F.A. Lloyd, Kuberan Pushparajah, Andrew P. King, Maria Deprez
Congenital Heart Disease (CHD) is a group of cardiac malformations present already during fetal life, representing the prevailing category of birth defects globally. Our aim in this study is to aid 3D fetal vessel topology visualisation in aortic arch anomalies, a group which encompasses a range of conditions with significant anatomical heterogeneity. We present a multi-task framework for automated multi-class fetal vessel segmentation from 3D black blood T2w MRI and anomaly classification. Our training data consists of binary manual segmentation masks of the cardiac vessels' region in individual subjects and fully-labelled anomaly-specific population atlases. Our framework combines deep learning label propagation using VoxelMorph with 3D Attention U-Net segmentation and DenseNet121 anomaly classification. We target 11 cardiac vessels and three distinct aortic arch anomalies, including double aortic arch, right aortic arch, and suspected coarctation of the aorta. We incorporate an anomaly classifier into our segmentation pipeline, delivering a multi-task framework with the primary motivation of correcting topological inaccuracies of the segmentation. The hypothesis is that the multi-task approach will encourage the segmenter network to learn anomaly-specific features. As a secondary motivation, an automated diagnosis tool may have the potential to enhance diagnostic confidence in a decision support setting. Our results showcase that our proposed training strategy significantly outperforms label propagation and a network trained exclusively on propagated labels. Our classifier outperforms a classifier trained exclusively on T2w volume images, with an average balanced accuracy of 0.99 (0.01) after joint training. Adding a classifier improves the anatomical and topological accuracy of all correctly classified double aortic arch subjects.
{"title":"Multi-task learning for joint weakly-supervised segmentation and aortic arch anomaly classification in fetal cardiac MRI","authors":"Paula Ramirez, Alena Uus, Milou P.M. van Poppel, Irina Grigorescu, Johannes K. Steinweg, David F.A. Lloyd, Kuberan Pushparajah, Andrew P. King, Maria Deprez","doi":"10.59275/j.melba.2023-b7bc","DOIUrl":"https://doi.org/10.59275/j.melba.2023-b7bc","url":null,"abstract":"Congenital Heart Disease (CHD) is a group of cardiac malformations present already during fetal life, representing the prevailing category of birth defects globally. Our aim in this study is to aid 3D fetal vessel topology visualisation in aortic arch anomalies, a group which encompasses a range of conditions with significant anatomical heterogeneity. We present a multi-task framework for automated multi-class fetal vessel segmentation from 3D black blood T2w MRI and anomaly classification. Our training data consists of binary manual segmentation masks of the cardiac vessels' region in individual subjects and fully-labelled anomaly-specific population atlases. Our framework combines deep learning label propagation using VoxelMorph with 3D Attention U-Net segmentation and DenseNet121 anomaly classification. We target 11 cardiac vessels and three distinct aortic arch anomalies, including double aortic arch, right aortic arch, and suspected coarctation of the aorta. We incorporate an anomaly classifier into our segmentation pipeline, delivering a multi-task framework with the primary motivation of correcting topological inaccuracies of the segmentation. The hypothesis is that the multi-task approach will encourage the segmenter network to learn anomaly-specific features. As a secondary motivation, an automated diagnosis tool may have the potential to enhance diagnostic confidence in a decision support setting. Our results showcase that our proposed training strategy significantly outperforms label propagation and a network trained exclusively on propagated labels. Our classifier outperforms a classifier trained exclusively on T2w volume images, with an average balanced accuracy of 0.99 (0.01) after joint training. Adding a classifier improves the anatomical and topological accuracy of all correctly classified double aortic arch subjects.","PeriodicalId":75083,"journal":{"name":"The journal of machine learning for biomedical imaging","volume":"2 3","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135041952","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-11-10DOI: 10.59275/j.melba.2023-3d9d
Abhineet Singh, Ila Jasra, Omar Mouhammed, Nidheesh Dadheech, Nilanjan Ray, James Shapiro
This paper presents advancements in automated early-stage prediction of the success of reprogramming human induced pluripotent stem cells (iPSCs) as a potential source for regenerative cell therapies. The minuscule success rate of iPSC-reprogramming of around 0.01% to 0.1% makes it labor-intensive, time-consuming, and exorbitantly expensive to generate a stable iPSC line since that requires culturing of millions of cells and intense biological scrutiny of multiple clones to identify a single optimal clone. The ability to reliably predict which cells are likely to establish as an optimal iPSC line at an early stage of pluripotency would therefore be ground-breaking in rendering this a practical and cost-effective approach to personalized medicine. Temporal information about changes in cellular appearance over time is crucial for predicting its future growth outcomes. In order to generate this data, we first performed continuous time-lapse imaging of iPSCs in culture using an ultra-high resolution microscope. We then annotated the locations and identities of cells in late-stage images where reliable manual identification is possible. Next, we propagated these labels backwards in time using a semi-automated tracking system to obtain labels for early stages of growth. Finally, we used this data to train deep neural networks to perform automatic cell segmentation and classification. Our code and data are available at https://github.com/abhineet123/ipsc_prediction
{"title":"Towards Early Prediction of Human iPSC Reprogramming Success","authors":"Abhineet Singh, Ila Jasra, Omar Mouhammed, Nidheesh Dadheech, Nilanjan Ray, James Shapiro","doi":"10.59275/j.melba.2023-3d9d","DOIUrl":"https://doi.org/10.59275/j.melba.2023-3d9d","url":null,"abstract":"This paper presents advancements in automated early-stage prediction of the success of reprogramming human induced pluripotent stem cells (iPSCs) as a potential source for regenerative cell therapies. The minuscule success rate of iPSC-reprogramming of around 0.01% to 0.1% makes it labor-intensive, time-consuming, and exorbitantly expensive to generate a stable iPSC line since that requires culturing of millions of cells and intense biological scrutiny of multiple clones to identify a single optimal clone. The ability to reliably predict which cells are likely to establish as an optimal iPSC line at an early stage of pluripotency would therefore be ground-breaking in rendering this a practical and cost-effective approach to personalized medicine.<br>Temporal information about changes in cellular appearance over time is crucial for predicting its future growth outcomes. In order to generate this data, we first performed continuous time-lapse imaging of iPSCs in culture using an ultra-high resolution microscope. We then annotated the locations and identities of cells in late-stage images where reliable manual identification is possible. Next, we propagated these labels backwards in time using a semi-automated tracking system to obtain labels for early stages of growth. Finally, we used this data to train deep neural networks to perform automatic cell segmentation and classification.<br>Our code and data are available at <a href='https://github.com/abhineet123/ipsc_prediction'>https://github.com/abhineet123/ipsc_prediction</a>","PeriodicalId":75083,"journal":{"name":"The journal of machine learning for biomedical imaging","volume":" 1282","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135186708","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-09-14DOI: 10.59275/j.melba.2023-219c
Dimitri Hamzaoui, Sarah Montagne, Raphaële Renard-Penna, Nicholas Ayache, Hervé Delingette
The extraction of consensus segmentations from several binary or probabilistic masks is important to solve various tasks such as the analysis of inter-rater variability or the fusion of several neural network outputs. One of the most widely used methods to obtain such a consensus segmentation is the STAPLE algorithm. In this paper, we first demonstrate that the output of that algorithm is heavily impacted by the background size of images and the choice of the prior. We then propose a new method to construct a binary or a probabilistic consensus segmentation based on the Fr'{e}chet means of carefully chosen distances which makes it totally independent of the image background size. We provide a heuristic approach to optimize this criterion such that a voxel's class is fully determined by its voxel-wise distance to the different masks, the connected component it belongs to and the group of raters who segmented it. We compared extensively our method on several datasets with the STAPLE method and the naive segmentation averaging method, showing that it leads to binary consensus masks of intermediate size between Majority Voting and STAPLE and to different posterior probabilities than Mask Averaging and STAPLE methods. Our code is available at https://gitlab.inria.fr/dhamzaou/jaccardmap .
{"title":"Morphologically-Aware Consensus Computation via Heuristics-based IterATive Optimization (MACCHIatO)","authors":"Dimitri Hamzaoui, Sarah Montagne, Raphaële Renard-Penna, Nicholas Ayache, Hervé Delingette","doi":"10.59275/j.melba.2023-219c","DOIUrl":"https://doi.org/10.59275/j.melba.2023-219c","url":null,"abstract":"The extraction of consensus segmentations from several binary or probabilistic masks is important to solve various tasks such as the analysis of inter-rater variability or the fusion of several neural network outputs. One of the most widely used methods to obtain such a consensus segmentation is the STAPLE algorithm. In this paper, we first demonstrate that the output of that algorithm is heavily impacted by the background size of images and the choice of the prior. We then propose a new method to construct a binary or a probabilistic consensus segmentation based on the Fr'{e}chet means of carefully chosen distances which makes it totally independent of the image background size. We provide a heuristic approach to optimize this criterion such that a voxel's class is fully determined by its voxel-wise distance to the different masks, the connected component it belongs to and the group of raters who segmented it. We compared extensively our method on several datasets with the STAPLE method and the naive segmentation averaging method, showing that it leads to binary consensus masks of intermediate size between Majority Voting and STAPLE and to different posterior probabilities than Mask Averaging and STAPLE methods. Our code is available at https://gitlab.inria.fr/dhamzaou/jaccardmap .","PeriodicalId":75083,"journal":{"name":"The journal of machine learning for biomedical imaging","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135551969","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-04-19DOI: 10.59275/j.melba.2023-18c1
Ashay Patel, Petru-Danial Tudiosu, Walter H.L. Pinaya, Gary Cook, Vicky Goh, Sebastien Ourselin, M. Jorge Cardoso
Cancer is a highly heterogeneous condition that can occur almost anywhere in the human body. [18F]fluorodeoxyglucose Positron Emission Tomography (18F-FDG PET) is a imaging modality commonly used to detect cancer due to its high sensitivity and clear visualisation of the pattern of metabolic activity. Nonetheless, as cancer is highly heterogeneous, it is challenging to train general-purpose discriminative cancer detection models, with data availability and disease complexity often cited as a limiting factor. Unsupervised learning methods, more specifically anomaly detection models, have been suggested as a putative solution. These models learn a healthy representation of tissue and detect cancer by predicting deviations from the healthy norm, which requires models capable of accurately learning long-range interactions between organs, their imaging patterns, and other abstract features with high levels of expressivity. Such characteristics are suitably satisfied by transformers, which have been shown to generate state-of-the-art results in unsupervised anomaly detection by training on normal data. This work expands upon such approaches by introducing multi-modal conditioning of the transformer via cross-attention i.e. supplying anatomical reference information from paired CT images to aid the PET anomaly detection task. Furthermore, we show the importance and impact of codebook sizing within a Vector Quantized Variational Autoencoder, on the ability of the transformer network to fulfill the task of anomaly detection. Using 294 whole-body PET/CT samples containing various cancer types, we show that our anomaly detection method is robust and capable of achieving accurate cancer localization results even in cases where normal training data is unavailable. In addition, we show the efficacy of this approach on out-of-sample data showcasing the generalizability of this approach even with limited training data. Lastly, we propose to combine model uncertainty with a new kernel density estimation approach, and show that it provides clinically and statistically significant improvements in accuracy and robustness, when compared to the classic residual-based anomaly maps. Overall, a superior performance is demonstrated against leading state-of-the-art alternatives, drawing attention to the potential of these approaches.
{"title":"Cross Attention Transformers for Multi-modal Unsupervised Whole-Body PET Anomaly Detection","authors":"Ashay Patel, Petru-Danial Tudiosu, Walter H.L. Pinaya, Gary Cook, Vicky Goh, Sebastien Ourselin, M. Jorge Cardoso","doi":"10.59275/j.melba.2023-18c1","DOIUrl":"https://doi.org/10.59275/j.melba.2023-18c1","url":null,"abstract":"Cancer is a highly heterogeneous condition that can occur almost anywhere in the human body. [<sup>18</sup>F]fluorodeoxyglucose Positron Emission Tomography (<sup>18</sup>F-FDG PET) is a imaging modality commonly used to detect cancer due to its high sensitivity and clear visualisation of the pattern of metabolic activity. Nonetheless, as cancer is highly heterogeneous, it is challenging to train general-purpose discriminative cancer detection models, with data availability and disease complexity often cited as a limiting factor. Unsupervised learning methods, more specifically anomaly detection models, have been suggested as a putative solution. These models learn a healthy representation of tissue and detect cancer by predicting deviations from the healthy norm, which requires models capable of accurately learning long-range interactions between organs, their imaging patterns, and other abstract features with high levels of expressivity. Such characteristics are suitably satisfied by transformers, which have been shown to generate state-of-the-art results in unsupervised anomaly detection by training on normal data. This work expands upon such approaches by introducing multi-modal conditioning of the transformer via cross-attention i.e. supplying anatomical reference information from paired CT images to aid the PET anomaly detection task. Furthermore, we show the importance and impact of codebook sizing within a Vector Quantized Variational Autoencoder, on the ability of the transformer network to fulfill the task of anomaly detection. Using 294 whole-body PET/CT samples containing various cancer types, we show that our anomaly detection method is robust and capable of achieving accurate cancer localization results even in cases where normal training data is unavailable. In addition, we show the efficacy of this approach on out-of-sample data showcasing the generalizability of this approach even with limited training data. Lastly, we propose to combine model uncertainty with a new kernel density estimation approach, and show that it provides clinically and statistically significant improvements in accuracy and robustness, when compared to the classic residual-based anomaly maps. Overall, a superior performance is demonstrated against leading state-of-the-art alternatives, drawing attention to the potential of these approaches.","PeriodicalId":75083,"journal":{"name":"The journal of machine learning for biomedical imaging","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135808044","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-02-17DOI: 10.59275/j.melba.2022-8g31
Zipei Zhao, Fengqian Pang, Yaou Liu, Zhiwen Liu, Chuyang Ye
Cell detection in histopathology images is of great interest to clinical practice and research, and convolutional neural networks (CNNs) have achieved remarkable cell detection results. Typically, to train CNN-based cell detection models, every positive instance in the training images needs to be annotated, and instances that are not labeled as positive are considered negative samples. However, manual cell annotation is complicated due to the large number and diversity of cells, and it can be difficult to ensure the annotation of every positive instance. In many cases, only incomplete annotations are available, where some of the positive instances are annotated and the others are not, and the classification loss term for negative samples in typical network training becomes incorrect. In this work, to address this problem of incomplete annotations, we propose to reformulate the training of the detection network as a positive-unlabeled learning problem. Since the instances in unannotated regions can be either positive or negative, they have unknown labels. Using the samples with unknown labels and the positively labeled samples, we first derive an approximation of the classification loss term corresponding to negative samples for binary cell detection, and based on this approximation we further extend the proposed framework to multi-class cell detection. For evaluation, experiments were performed on four publicly available datasets. The experimental results show that our method improves the performance of cell detection in histopathology images given incomplete annotations for network training.
{"title":"Positive-unlabeled learning for binary and multi-class cell detection in histopathology images with incomplete annotations","authors":"Zipei Zhao, Fengqian Pang, Yaou Liu, Zhiwen Liu, Chuyang Ye","doi":"10.59275/j.melba.2022-8g31","DOIUrl":"https://doi.org/10.59275/j.melba.2022-8g31","url":null,"abstract":"Cell detection in histopathology images is of great interest to clinical practice and research, and convolutional neural networks (CNNs) have achieved remarkable cell detection results. Typically, to train CNN-based cell detection models, every positive instance in the training images needs to be annotated, and instances that are not labeled as positive are considered negative samples. However, manual cell annotation is complicated due to the large number and diversity of cells, and it can be difficult to ensure the annotation of every positive instance. In many cases, only incomplete annotations are available, where some of the positive instances are annotated and the others are not, and the classification loss term for negative samples in typical network training becomes incorrect. In this work, to address this problem of incomplete annotations, we propose to reformulate the training of the detection network as a positive-unlabeled learning problem. Since the instances in unannotated regions can be either positive or negative, they have unknown labels. Using the samples with unknown labels and the positively labeled samples, we first derive an approximation of the classification loss term corresponding to negative samples for binary cell detection, and based on this approximation we further extend the proposed framework to multi-class cell detection. For evaluation, experiments were performed on four publicly available datasets. The experimental results show that our method improves the performance of cell detection in histopathology images given incomplete annotations for network training.","PeriodicalId":75083,"journal":{"name":"The journal of machine learning for biomedical imaging","volume":"89 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135339724","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-01-10DOI: 10.59275/j.melba.2022-ac28
A. Zeidan, Paula Ramirez Gilliland, Ashay Patel, Zhanchong Ou, Dimitra Flouri, N. Mufti, K. Maksym, Rosalind Aughwane, S. Ourselin, Anna L. David, A. Melbourne
Fetal growth restriction (FGR) is a prevalent pregnancy condition characterised by failure of the fetus to reach its genetically predetermined growth potential. The multiple aetiologies, coupled with the risk of fetal complications - encompassing neurodevelopmental delay, neonatal morbidity, and stillbirth - motivate the need to improve holistic assessment of the FGR fetus using MRI. We hypothesised that the fetal liver and placenta would provide insights into FGR biomarkers, unattainable through conventional methods. Therefore, we explore the application of model fitting techniques, linear regression machine learning models, deep learning regression, and Haralick textured features from multi-contrast MRI for multi-fetal organ analysis of FGR. We employed T2 relaxometry and diffusion-weighted MRI datasets (using a combined T2-diffusion scan) for 12 normally grown and 12 FGR gestational age (GA) matched pregnancies (Estimated Fetal Weight below 3rd centile, Median 28+/-3wks). We applied the Intravoxel Incoherent Motion Model, which describes circulatory properties of the fetal organs, and analysed the resulting features distinguishing both cohorts. We additionally used novel multi-compartment models for MRI fetal analysis, which exhibit potential to provide a multi-organ FGR assessment, overcoming the limitations of empirical indicators - such as abnormal artery Doppler findings - to evaluate placental dysfunction. The placenta and fetal liver presented key differentiators between FGR and normal controls, with significant decreased perfusion, abnormal fetal blood motion and reduced fetal blood oxygenation. This may be associated with the preferential shunting of the fetal blood towards the fetal brain, affecting supply to the liver. These features were further explored to determine their role in assessing FGR severity, by employing simple machine learning models to predict FGR diagnosis (100% accuracy in test data, n=5), GA at delivery, time from MRI scan to delivery, and baby weight. We additionally explored the use of deep learning to regress the latter three variables, training a convolutional neural network with our liver and placenta voxel-level parameter maps, obtained from our multi-compartment model fitting. Image texture analysis of the fetal organs demonstrated prominent textural variations in the placental perfusion fractions maps between the groups (p<0.0009), and spatial differences in the incoherent fetal capillary blood motion in the liver (p<0.009). This research serves as a proof-of-concept, investigating the effect of FGR on fetal organs, measuring differences in perfusion and oxygenation within the placenta and fetal liver, and their prognostic importance in automated diagnosis using simple machine learning models.
{"title":"An Approach to Automated Diagnosis and Texture Analysis of the Fetal Liver & Placenta in Fetal Growth Restriction","authors":"A. Zeidan, Paula Ramirez Gilliland, Ashay Patel, Zhanchong Ou, Dimitra Flouri, N. Mufti, K. Maksym, Rosalind Aughwane, S. Ourselin, Anna L. David, A. Melbourne","doi":"10.59275/j.melba.2022-ac28","DOIUrl":"https://doi.org/10.59275/j.melba.2022-ac28","url":null,"abstract":"Fetal growth restriction (FGR) is a prevalent pregnancy condition characterised by failure of the fetus to reach its genetically predetermined growth potential. The multiple aetiologies, coupled with the risk of fetal complications - encompassing neurodevelopmental delay, neonatal morbidity, and stillbirth - motivate the need to improve holistic assessment of the FGR fetus using MRI. We hypothesised that the fetal liver and placenta would provide insights into FGR biomarkers, unattainable through conventional methods. Therefore, we explore the application of model fitting techniques, linear regression machine learning models, deep learning regression, and Haralick textured features from multi-contrast MRI for multi-fetal organ analysis of FGR. We employed T2 relaxometry and diffusion-weighted MRI datasets (using a combined T2-diffusion scan) for 12 normally grown and 12 FGR gestational age (GA) matched pregnancies (Estimated Fetal Weight below 3rd centile, Median 28+/-3wks). We applied the Intravoxel Incoherent Motion Model, which describes circulatory properties of the fetal organs, and analysed the resulting features distinguishing both cohorts. We additionally used novel multi-compartment models for MRI fetal analysis, which exhibit potential to provide a multi-organ FGR assessment, overcoming the limitations of empirical indicators - such as abnormal artery Doppler findings - to evaluate placental dysfunction. The placenta and fetal liver presented key differentiators between FGR and normal controls, with significant decreased perfusion, abnormal fetal blood motion and reduced fetal blood oxygenation. This may be associated with the preferential shunting of the fetal blood towards the fetal brain, affecting supply to the liver. These features were further explored to determine their role in assessing FGR severity, by employing simple machine learning models to predict FGR diagnosis (100% accuracy in test data, n=5), GA at delivery, time from MRI scan to delivery, and baby weight. We additionally explored the use of deep learning to regress the latter three variables, training a convolutional neural network with our liver and placenta voxel-level parameter maps, obtained from our multi-compartment model fitting. Image texture analysis of the fetal organs demonstrated prominent textural variations in the placental perfusion fractions maps between the groups (p<0.0009), and spatial differences in the incoherent fetal capillary blood motion in the liver (p<0.009). This research serves as a proof-of-concept, investigating the effect of FGR on fetal organs, measuring differences in perfusion and oxygenation within the placenta and fetal liver, and their prognostic importance in automated diagnosis using simple machine learning models.","PeriodicalId":75083,"journal":{"name":"The journal of machine learning for biomedical imaging","volume":"16 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81025448","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-15DOI: 10.59275/j.melba.2023-8172
Alban Gossard, F. de Gournay, P. Weiss
Data-driven optimization of sampling patterns in MRI has recently received a significant attention. Following recent observations on the combinatorial number of minimizers in off-the-grid optimization, we propose a framework to globally optimize the sampling densities using Bayesian optimization. Using a dimension reduction technique, we optimize the sampling trajectories more than 20 times faster than conventional off-the-grid methods, with a restricted number of training samples. This method – among other benefits – discards the need of automatic differentiation. Its performance is slightly worse than state-of-the-art learned trajectories since it reduces the space of admissible trajectories, but comes with significant computational advantages. Other contributions include: i) a careful evaluation of the distance in probability space to generate trajectories ii) a specific training procedure on families of operators for unrolled reconstruction networks and iii) a gradient projection based scheme for trajectory optimization.
{"title":"Bayesian Optimization of Sampling Densities in MRI","authors":"Alban Gossard, F. de Gournay, P. Weiss","doi":"10.59275/j.melba.2023-8172","DOIUrl":"https://doi.org/10.59275/j.melba.2023-8172","url":null,"abstract":"Data-driven optimization of sampling patterns in MRI has recently received a significant attention. Following recent observations on the combinatorial number of minimizers in off-the-grid optimization, we propose a framework to globally optimize the sampling densities using Bayesian optimization. Using a dimension reduction technique, we optimize the sampling trajectories more than 20 times faster than conventional off-the-grid methods, with a restricted number of training samples. This method – among other benefits – discards the need of automatic differentiation. Its performance is slightly worse than state-of-the-art learned trajectories since it reduces the space of admissible trajectories, but comes with significant computational advantages. Other contributions include: i) a careful evaluation of the distance in probability space to generate trajectories ii) a specific training procedure on families of operators for unrolled reconstruction networks and iii) a gradient projection based scheme for trajectory optimization.","PeriodicalId":75083,"journal":{"name":"The journal of machine learning for biomedical imaging","volume":"55 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2022-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90322877","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-17DOI: 10.59275/j.melba.2022-9e4b
S. Ellis, O. M. Manzanera, V. Baltatzis, Ibrahim Nawaz, A. Nair, L. L. Folgoc, S. Desai, Ben Glocker, J. Schnabel
Generative adversarial networks (GANs) are able to model accurately the distribution of complex, high-dimensional datasets, for example images. This characteristic makes high-quality GANs useful for unsupervised anomaly detection in medical imaging. However, differences in training datasets such as output image dimensionality and appearance of semantically meaningful features mean that GAN models from the natural image processing domain may not work 'out-of-the-box' for medical imaging applications, necessitating re-implementation and re-evaluation. In this work we adapt and evaluate three GAN models to the application of modelling 3D healthy image patches for pulmonary CT. To the best of our knowledge, this is the first time that such a detailed evaluation has been performed. The deep convolutional GAN (DCGAN), styleGAN and the bigGAN architectures were selected for investigation due to their ubiquity and high performance in natural image processing. We train different variants of these methods and assess their performance using the widely used Frechet Inception Distance (FID). In addition, the quality of the generated images was evaluated by a human observer study, the ability of the networks to model 3D domain-specific features was investigated, and the structure of the GAN latent spaces was analysed. Results show that the 3D styleGAN approaches produce realistic-looking images with meaningful 3D structure, but suffer from mode collapse which must be explicitly addressed during training to obtain diversity in the samples. Conversely, the 3D DCGAN models show a greater capacity for image variability, but at the cost of poor-quality images. The 3D bigGAN models provide an intermediate level of image quality, but most accurately model the distribution of selected semantically meaningful features. The results suggest that future development is required to realise a 3D GAN with sufficient representational capacity for patch-based lung CT anomaly detection and we offer recommendations for future areas of research, such as experimenting with other architectures and incorporation of position-encoding.
{"title":"Evaluation of 3D GANs for Lung Tissue Modelling in Pulmonary CT","authors":"S. Ellis, O. M. Manzanera, V. Baltatzis, Ibrahim Nawaz, A. Nair, L. L. Folgoc, S. Desai, Ben Glocker, J. Schnabel","doi":"10.59275/j.melba.2022-9e4b","DOIUrl":"https://doi.org/10.59275/j.melba.2022-9e4b","url":null,"abstract":"Generative adversarial networks (GANs) are able to model accurately the distribution of complex, high-dimensional datasets, for example images. This characteristic makes high-quality GANs useful for unsupervised anomaly detection in medical imaging. However, differences in training datasets such as output image dimensionality and appearance of semantically meaningful features mean that GAN models from the natural image processing domain may not work 'out-of-the-box' for medical imaging applications, necessitating re-implementation and re-evaluation. In this work we adapt and evaluate three GAN models to the application of modelling 3D healthy image patches for pulmonary CT. To the best of our knowledge, this is the first time that such a detailed evaluation has been performed. The deep convolutional GAN (DCGAN), styleGAN and the bigGAN architectures were selected for investigation due to their ubiquity and high performance in natural image processing. We train different variants of these methods and assess their performance using the widely used Frechet Inception Distance (FID). In addition, the quality of the generated images was evaluated by a human observer study, the ability of the networks to model 3D domain-specific features was investigated, and the structure of the GAN latent spaces was analysed. Results show that the 3D styleGAN approaches produce realistic-looking images with meaningful 3D structure, but suffer from mode collapse which must be explicitly addressed during training to obtain diversity in the samples. Conversely, the 3D DCGAN models show a greater capacity for image variability, but at the cost of poor-quality images. The 3D bigGAN models provide an intermediate level of image quality, but most accurately model the distribution of selected semantically meaningful features. The results suggest that future development is required to realise a 3D GAN with sufficient representational capacity for patch-based lung CT anomaly detection and we offer recommendations for future areas of research, such as experimenting with other architectures and incorporation of position-encoding.","PeriodicalId":75083,"journal":{"name":"The journal of machine learning for biomedical imaging","volume":"23 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2022-08-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82734864","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号