Journal of Medical Imaging最新文献

Purpose: Accurate cancer subtyping is essential for precision medicine but challenged by the computational demands of gigapixel whole-slide images (WSIs). Although transformer-based multiple instance learning (MIL) methods achieve strong performance, their quadratic complexity limits clinical deployment. We introduce LiteMIL, a computationally efficient cross-attention MIL, optimized for WSIs classification.

Approach: LiteMIL employs a single learnable query with multi-head cross-attention for bag-level aggregation from extracted features. We evaluated LiteMIL against five baselines (mean/max pooling, ABMIL, MAD-MIL, and TransMIL) on four TCGA datasets (breast: $n=875$ , kidney: $n=906$ , lung: $n=958$ , and TUPAC16: $n=821$ ) using nested cross-validation with patient-level splitting. Systematic ablation studies evaluated multi-query variants, attention heads, dropout rates, and architectural components.

Results: LiteMIL achieved competitive accuracy (average 83.5%), matching TransMIL, while offering substantial efficiency gains: $4.8\times$ fewer parameters (560K versus 2.67M), $2.9\times$ faster inference (1.6s versus 4.6s per fold), and $6.7\times$ lower Graphics Processing Unit (GPU) memory usage (1.15 GB versus 7.77 GB). LiteMIL excelled on lung (86.3% versus 85.0%), TUPAC16 (72% versus 71.4%), and matched kidney performance (89.9% versus 89.7%). Ablation studies revealed task-dependent multi-query performance benefits: $Q=4$ versus $Q=1$ improved morphologically heterogeneous tasks (breast/lung $+1.3\%$ each, $p<0.05$ ) but degraded on grading tasks (TUPAC16: $-1.6\%$ ), validating single-query optimality for focused attention scenarios.

Conclusions: LiteMIL provides a resource-efficient solution for WSI classification. The cross-attention architecture matches complex transformer performance while enabling deployment on consumer GPUs. Task-dependent design insights, single query for sparse discriminating features, multi-query for heterogeneous patterns, guide practical implementation. The architecture's efficiency, combined with compact features, makes LiteMIL suitable for clinical integration in settings with limited computational infrastructure.

Purpose: Centralized machine learning often struggles with limited data access and expert involvement. We investigate decentralized approaches that preserve data privacy while enabling collaborative model training for medical imaging tasks.

Approach: We explore asynchronous federated learning (FL) using the FL with buffered asynchronous aggregation (FedBuff) algorithm for classifying optical coherence tomography (OCT) retina images. Unlike synchronous algorithms such as FedAvg, which require all clients to participate simultaneously, FedBuff supports independent client updates. We compare its performance to both centralized models and FedAvg. In addition, we develop a browser-based proof-of-concept system using modern web technologies to assess the feasibility and limitations of interactive, collaborative learning in real-world settings.

Results: FedBuff performs well in binary OCT classification tasks but shows reduced accuracy in more complex, multiclass scenarios. FedAvg achieves results comparable to centralized training, consistent with previous findings. Although FedBuff underperforms compared with FedAvg and centralized models, it still delivers acceptable accuracy in less complex settings. The browser-based prototype demonstrates the potential for accessible, user-driven FL systems but also highlights technical limitations in current web standards, especially regarding local computation and communication efficiency.

Conclusion: Asynchronous FL via FedBuff offers a promising, privacy-preserving approach for medical image classification, particularly when synchronous participation is impractical. However, its scalability to complex classification tasks remains limited. Web-based implementations have the potential to broaden access to collaborative AI tools, but limitations of the current technologies need to be further investigated.

Purpose: There are multiple commercially available, Food and Drug Administration (FDA)-cleared, artificial intelligence (AI)-based tools automating stroke evaluation in noncontrast computed tomography (NCCT). This study assessed the impact of variations in reconstruction kernel and slice thickness on two outputs of such a system: hypodense volume and Alberta Stroke Program Early CT Score (ASPECTS).

Approach: The NCCT series image data of 67 patients imaged with a CT stroke protocol were reconstructed with four kernels (H10s-smooth, H40s-medium, H60s-sharp, and H70h-very sharp) and three slice thicknesses (1.5, 3.0, and 5.0 mm) to create 1 reference condition (H40s/5.0 mm) and 11 nonreference conditions. The 12 reconstructions per patient were processed with a commercially available FDA-cleared software package that yields total hypodense volume (mL) and ASPECTS. A mixed-effect model was used to test the difference in hypodense volume, and an ordered logistic model was used to test the difference in e-ASPECTS.

Results: Hypodense volume differences from the reference condition ranged from $-14.6$ to 1.1 mL and were significant for all nonreference kernels (H10s $p=0.025$ , H60s $p<0.001$ , and H70h $p<0.001$ ) and for thinner slices (1.5 mm $p<0.001$ and 3.0 mm $p=0.002$ ). e-ASPECTS was invariant to the nonreference kernels and slice thicknesses, with a mean difference ranging from $-0.1$ to 0.5. No significant differences were found for any kernel or slice thickness (all $p>0.05$ ).

Conclusions: Automated hypodense volume measured with a commercially available, FDA-cleared software package is substantially impacted by reconstruction kernel and slice thickness. Conversely, automated ASPECTS is invariant to these reconstruction parameters.

Purpose: Cochlear implant (CI) surgery treats severe hearing loss by inserting an electrode array into the cochlea to stimulate the auditory nerve. An important step in this procedure is mastoidectomy, which removes part of the mastoid region of the temporal bone to provide surgical access. Accurate mastoidectomy shape prediction from preoperative imaging improves presurgical planning, reduces risks, and enhances surgical outcomes. Despite its importance, there are limited deep-learning-based studies regarding this topic due to the challenges of acquiring ground-truth labels. We address this gap by investigating self-supervised and weakly-supervised learning models to predict the mastoidectomy region without human annotations.

Approach: We propose a hybrid self-supervised and weakly-supervised learning framework to predict the mastoidectomy region directly from preoperative CT scans, where the mastoid remains intact. Our self-supervised learning approach reconstructs the postmastoidectomy 3D surface from preoperative imaging, aiming to align with the corresponding intraoperative microscope views for future surgical navigation-related applications. Postoperative CT scans are used in the self-supervised learning model to assist training procedures despite additional challenges such as metal artifacts and low signal-to-noise ratios introduced by them. To further improve the accuracy and robustness, we introduce a Mamba-based weakly-supervised model that refines mastoidectomy shape prediction by using 3D T-distribution loss function, inspired by the student- $t$ distribution. Weak supervision is achieved by leveraging segmentation results from the prior self-supervised framework, eliminating the manual data labeling process.

Results: Our hybrid method achieves a mean Dice score of 0.72 when predicting the complex and boundary-less mastoidectomy shape, surpassing state-of-the-art approaches and demonstrating strong performance. The method provides groundwork for constructing 3D postmastoidectomy surfaces directly from the corresponding preoperative CT scans.

Conclusion: To our knowledge, this is the first work that integrates self-supervised and weakly-supervised learning for mastoidectomy shape prediction, offering a robust and efficient solution for CI surgical planning while leveraging 3D T-distribution loss in weakly-supervised medical imaging.

Purpose: Accurate simulation of breast tissue deformation is essential for reliable image registration between 3D imaging modalities and 2D mammograms, where compression significantly alters tissue geometry. Although finite element analysis (FEA) provides high-fidelity modeling, it is computationally intensive and not well suited for rapid simulations. To address this, the physics-based graph neural network (PhysGNN) has been introduced as a computationally efficient approximation model trained on FEA-generated deformations. We extend prior work by evaluating the performance of PhysGNN on new digital breast phantoms and assessing the impact of training on multiple phantoms.

Approach: PhysGNN was trained on both single-phantom (per-geometry) and multiphantom (multigeometry) datasets generated from incremental FEA simulations. The digital breast phantoms represent the uncompressed state, serving as input geometries for predicting compressed configurations. A leave-one-deformation-out evaluation strategy was used to assess predictive performance under compression.

Results: Training on new digital phantoms confirmed the model's robust performance, though with some variability in prediction accuracy reflecting the diverse anatomical structures. Multiphantom training further enhanced this robustness and reduced prediction errors.

Conclusions: PhysGNN offers a computationally efficient alternative to FEA for simulating breast compression. The results showed that model performance remains robust when trained per-geometry, and further demonstrated that multigeometry training enhances predictive accuracy and robustness for the geometries included in the training set. This suggests a strong potential path toward developing reliable models for generating compressed breast volumes, which could facilitate image registration and algorithm development.

Purpose: Thermal ablation is a minimally invasive therapy used for the treatment of small renal cell carcinoma tumors. Treatment success is evaluated on postablation computed tomography (CT) to determine if the ablation zone covered the tumor with an adequate treatment margin (often 5 to 10 mm). Incorrect margin identification can lead to treatment misassessment, resulting in unnecessary additional ablation. Therefore, segmentation of the renal ablation zone (RAZ) is crucial for treatment evaluation. We aim to develop and assess an accurate deep learning workflow for delineating the RAZ from surrounding tissues in kidney CT images.

Approach: We present an advanced deep learning method using the attention-based U-Net architecture to segment the RAZ. The workflow leverages the strengths of U-Net, enhanced with attention mechanisms, to improve the network's focus on the most relevant parts of the images, resulting in an accurate segmentation.

Results: Our model was trained and evaluated on a dataset comprising 76 patients' annotated RAZs in CT images. Analysis demonstrated that the proposed workflow achieved an accuracy $=0.97\pm 0.02$ , precision $=0.74\pm 0.23$ , $\text{recall}=0.73\pm 0.25$ , $\mathrm{DSC}=0.70\pm 0.22$ , Jaccard $=0.58\pm 0.22$ , specificity $=0.99\pm 0.01$ , Hausdorff distance $=6.70\pm 4.44\mathrm{mm}$ , and mean absolute boundary distance $=2.67\pm 2.22\mathrm{mm}$ .

Conclusions: We used 3D CT images with RAZs and, for the first time, addressed deep-learning-based RAZ segmentation using parallel CT images. Our framework can effectively segment RAZs, allowing clinicians to automatically determine the ablation margin, making our tool ready for clinical use. Prediction time is $\sim 1s$ per patient, enabling clinicians to perform quick reviews, especially in time-constrained settings.

Purpose: Accurate airway measurement is critical for bronchitis quantification with computed tomography (CT), yet optimal protocols and the added value of photon-counting CT (PCCT) over energy-integrating CT (EICT) for reducing bias remain unclear. We quantified biomarker accuracy across modalities and protocols and assessed strategies to reduce bias.

Approach: A virtual imaging trial with 20 bronchitis anthropomorphic models was scanned using a validated simulator for two systems (EICT: SOMATOM Flash; PCCT: NAEOTOM Alpha) at 6.3 and 12.6 mGy. Reconstructions varied algorithm, kernel sharpness, slice thickness, and pixel size. Pi10 (square-root wall thickness at 10-mm perimeter) and WA% (wall-area percentage) were compared against ground-truth airway dimensions obtained from the 0.1-mm-precision anatomical models prior to CT simulation. External validation used clinical PCCT ( $n=22$ ) and EICT ( $n=80$ ).

Results: Simulated airway dimensions agreed with pathological references ( $R=0.89-0.93$ ). PCCT had lower errors than EICT across segmented generations ( $p<0.05$ ). Under optimal parameters, PCCT improved Pi10 and WA% accuracy by 26.3% and 64.9%. Across the tested PCCT and EICT imaging protocols, improvements were associated with sharper kernels (25.8% Pi10, 33.0% WA%), thinner slices (23.9% Pi10, 49.8% WA%), smaller pixels (17.0% Pi10, 23.1% WA%), and higher dose ( $\le 3.9\%$ ). Clinically, PCCT achieved higher maximum airway generation ( $8.8\pm 0.5$ versus $6.0\pm 1.1$ ) and lower variability, mirroring trends in virtual results.

Conclusions: PCCT improves the accuracy and consistency of airway biomarker quantification relative to EICT, particularly with optimized protocols. The validated virtual platform enables modality-bias assessment and protocol optimization for accurate, reproducible bronchitis measurements.

Purpose: Research in deep learning has shown a great advancement in the detection of melanoma. However, recent literature has emphasized a tendency of certain models to rely on disease-irrelevant visual artifacts such as dark corners, dense hair, or ruler marks. The dependence on these markers leads to biased models that do well for training but generalize poorly to heterogeneous clinical environments. To address these limitations in developing reliability in skin lesion detection, a lightweight cross-attention-based semantic dual (LCSD) transformer model was proposed.

Approach: The LCSD model extracts global-level semantic information, uses feature normalization to improve model accuracy, and employs semantic queries to improve domain generalization. Multihead attention is included with the semantic queries to refine global features. The cross-attention between feature maps and semantic query provides the model with a generalized encoding of the global context. The model improved the computational complexity from $O\left(n^{2}d\right)$ to $O(nmd+m^{2}d)$ , which makes the model suitable for the development of real-time and mobile applications.

Results: Empirical evaluation was conducted on three challenging datasets: Derm7pt-Dermoscopic, Derm7pt-Clinical, and PAD-UFES-20. The proposed model achieved classification accuracies of 82.82%, 72.95%, and 86.21%, respectively. These results demonstrate superior performance compared with conventional transformer-based models, highlighting both improved robustness and reduced computational cost.

Conclusion: The LCSD model mitigates the influence of irrelevant visual characteristics, enhances domain generalization, and ensures better adaptability across diverse clinical scenarios. Its lightweight design further supports deployment in mobile applications, making it a reliable and efficient solution for real-world melanoma detection.

Purpose: Colorectal cancer is the third most common cancer globally, with a high mortality rate due to metastatic progression, particularly in the liver. Surgical resection remains the main curative treatment, but only a small subset of patients is eligible for surgery at diagnosis. For patients with initially unresectable colorectal liver metastases (CRLM), neoadjuvant chemotherapy can downstage tumors, potentially making surgery feasible. We investigate whether radiomic signatures-quantitative imaging biomarkers derived from baseline computed tomography (CT) scans-can noninvasively predict chemotherapy response in patients with unresectable CRLM, offering a pathway toward personalized treatment planning.

Approach: We used radiomics combined with a stacking classifier (SC) to predict treatment outcome. Baseline CT imaging data from 355 patients with initially unresectable CRLM were analyzed using two regions of interest (ROIs) separately (all tumors in the liver and the largest tumor by volume). From each ROI, 107 radiomic features were extracted. The dataset was split into training and testing sets, and multiple machine learning models were trained and integrated via stacking to enhance prediction. Logistic regression coefficients were used to derive radiomic signatures.

Results: The SC achieved strong predictive performance, with an area under the receiver operating characteristic curve of up to 0.77 for response prediction. Logistic regression identified 12 and 7 predictive features for treatment response in all tumors and the largest tumor ROIs, respectively.

Conclusion: Our findings demonstrate that radiomic features from baseline CT scans can serve as robust, interpretable biomarkers for predicting chemotherapy response, offering insights to guide personalized treatment in unresectable CRLM.

Purpose: Triple-negative breast cancer (TNBC) is an aggressive subtype with limited treatment options and high recurrence rates. Magnetic resonance imaging (MRI) is widely used for tumor assessment, but manual segmentation is labor-intensive and variable. Existing deep learning methods often lack generalizability, calibrated confidence, and robust uncertainty quantification.

Approach: We propose ER²Net, an evidential reasoning-enabled neural network for reliable TNBC tumor segmentation on MRI. ER²Net trains multiple U-Net variants with dropouts to generate diverse predictions and introduces pixel-wise reliability to quantify model agreement. We then introduce two ensemble fusion techniques: weighted reliability (WR) segmentation, which leverages pixel-wise reliability to enhance sensitivity, and Bayesian fusion (BF), which integrates predictions probabilistically for robust consensus. Confidence calibration is achieved using evidential reasoning, and we further propose pixel-wise reliable confidence entropy (PWRE) as a uncertainty measure.

Results: ER²Net improved performance compared with individual models. WR achieved IoU = 0.886, sensitivity = 0.928, precision = 0.952, and Hausdorff distance = 5.429 mm, whereas BF achieved IoU = 0.885 and sensitivity = 0.929. Reliable fusion provided the best calibration [expected calibration error = 0.00003; maximum calibration error = 0.017]. PWRE produced lower variance than conventional entropy, yielding more stable uncertainty estimates.

Conclusion: ER²Net introduces WR segmentation and BF as enhanced fusion techniques and PWRE as a uncertainty metric. Together, these advances improve segmentation accuracy, sensitivity, confidence calibration, and uncertainty estimation, paving the way for reliable MRI-based tools to support personalized treatment planning and response assessment in TNBC.