Abstract Congenital anomalies of inferior vena cava are increasingly being recognized with the technical advancements and increased utilization of cross-sectional imaging techniques. Duplication of inferior vena cava classically involves duplication of the infrarenal segment, where both inferior vena cava ascend on either side of the abdominal aorta until they form a confluence at the level of the renal veins. It has been extensively described in literature with few reports of more complex variation in the form of duplicated infrarenal inferior vena cava with azygos or hemiazygos continuation. This article describes extremely rare complete duplication of inferior vena cava involving both suprarenal and infrarenal segments. Moreover, the complete duplication of inferior vena cava is seen in association with concomitant double superior vena cava, in a patient with visceroatrial situs solitus and associated congenital heart disease, which to the best of our knowledge, has not been reported so far in literature. This study also highlights the utility of multidetector computed tomography in accurate identification of such anomalies.
{"title":"Complete Duplication of Inferior Vena Cava Coexisting with Double Superior Vena Cava In Situ Solitus: Hitherto Unreported Pattern","authors":"Harsimran Bhatia, Arun Sharma, Sanjeev Hanumantacharya Naganur, Manphool Singhal","doi":"10.1055/s-0043-1775796","DOIUrl":"https://doi.org/10.1055/s-0043-1775796","url":null,"abstract":"Abstract Congenital anomalies of inferior vena cava are increasingly being recognized with the technical advancements and increased utilization of cross-sectional imaging techniques. Duplication of inferior vena cava classically involves duplication of the infrarenal segment, where both inferior vena cava ascend on either side of the abdominal aorta until they form a confluence at the level of the renal veins. It has been extensively described in literature with few reports of more complex variation in the form of duplicated infrarenal inferior vena cava with azygos or hemiazygos continuation. This article describes extremely rare complete duplication of inferior vena cava involving both suprarenal and infrarenal segments. Moreover, the complete duplication of inferior vena cava is seen in association with concomitant double superior vena cava, in a patient with visceroatrial situs solitus and associated congenital heart disease, which to the best of our knowledge, has not been reported so far in literature. This study also highlights the utility of multidetector computed tomography in accurate identification of such anomalies.","PeriodicalId":51597,"journal":{"name":"Indian Journal of Radiology and Imaging","volume":"217 5","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"136316811","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}
Abstract Perivascular epithelioid cell tumors (PEComas) are very rare mesenchymal tumors. In addition to being rare in general, they are even rarer in the liver. There are various subtypes and there is only one case of liver PEComa reported in the clear cell type in the literature to date. We present the second clear cell type liver PEComa in this case by reviewing the literature data. The hypodense, smooth-edged, ovoid lesion was detected on computed tomography (CT) for performing abdominal pain in a 41-year-old female patient. Magnetic resonance imaging (MRI) was then performed for lesion characterization. The lesion was hypointense on T1-weighted imaging (T1WI), hyperintense on T2-weighted imaging (T2WI). In dynamic phases, it showed marked enhancement on the arterial phase and capsular enhancement with central washout on the portal and late venous phases. The posterior branch of the right portal vein extended into the mass. The lesion was excised and the pathological result was epithelioid clear cell subtype of PEComa. Although the imaging findings are generally nonspecific and certain diagnosis is made histopathologically, radiologists should consider PEComa in the differential diagnosis in the presence of intensely enhanced lesion on the right lobe in female patients. Also, the “large vessel sign” may help in the diagnosis.
{"title":"Very Rare Liver Tumor: PEComa Case Report with and a Review of Literature","authors":"Celal Yazıcı, Elif Gündoğdu","doi":"10.1055/s-0043-1774392","DOIUrl":"https://doi.org/10.1055/s-0043-1774392","url":null,"abstract":"Abstract Perivascular epithelioid cell tumors (PEComas) are very rare mesenchymal tumors. In addition to being rare in general, they are even rarer in the liver. There are various subtypes and there is only one case of liver PEComa reported in the clear cell type in the literature to date. We present the second clear cell type liver PEComa in this case by reviewing the literature data. The hypodense, smooth-edged, ovoid lesion was detected on computed tomography (CT) for performing abdominal pain in a 41-year-old female patient. Magnetic resonance imaging (MRI) was then performed for lesion characterization. The lesion was hypointense on T1-weighted imaging (T1WI), hyperintense on T2-weighted imaging (T2WI). In dynamic phases, it showed marked enhancement on the arterial phase and capsular enhancement with central washout on the portal and late venous phases. The posterior branch of the right portal vein extended into the mass. The lesion was excised and the pathological result was epithelioid clear cell subtype of PEComa. Although the imaging findings are generally nonspecific and certain diagnosis is made histopathologically, radiologists should consider PEComa in the differential diagnosis in the presence of intensely enhanced lesion on the right lobe in female patients. Also, the “large vessel sign” may help in the diagnosis.","PeriodicalId":51597,"journal":{"name":"Indian Journal of Radiology and Imaging","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"136317221","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}
In this letter to the editor, we would like to report a case of incidental connatal cyst in a 50-year-old male patient. The patient was on migraine medications; however, he had recurrent episodes of headaches with nausea and was asked to perform a magnetic resonance imaging (MRI) by a treating physician.Thereisnopasthistoryofseizures,paresisorplegia, sensorineural de fi cits, cranial nerve involvement, urinary incontinence, etc. in the patient. The patient was born of a normal delivery at term, with no history of neonatal hospitalization, prolonged jaundice, and respiratory distress. Neurological examination was normal. On MRI, T2-weighted (T2W) coronal images showed bilateral hyperintense cysts located superolaterally to frontal horns ( ► Fig. 1A and B ). Similarly axialT2Wimagesshowedthehyperintensecystanteriortothe frontal horns ( ► Fig. 1C ). On fl uid-attenuated inversion recovery coronal images and T1W, they appeared hypointense ( ► Fig. 2A and B ) without any evidence of associated in fl am-mation and periventricular gliosis. Given the location and no signi fi cantneonatalhistoryinthepatients,the fi ndingsonMRI are consistent with frontal horn cyst aka coarctation of lateral ventricles or connatal cysts. Connatal
{"title":"Connatal Cyst in a 50-Year-Old Patient","authors":"Gayatri Harshe, Krishna Mundada","doi":"10.1055/s-0043-1775813","DOIUrl":"https://doi.org/10.1055/s-0043-1775813","url":null,"abstract":"In this letter to the editor, we would like to report a case of incidental connatal cyst in a 50-year-old male patient. The patient was on migraine medications; however, he had recurrent episodes of headaches with nausea and was asked to perform a magnetic resonance imaging (MRI) by a treating physician.Thereisnopasthistoryofseizures,paresisorplegia, sensorineural de fi cits, cranial nerve involvement, urinary incontinence, etc. in the patient. The patient was born of a normal delivery at term, with no history of neonatal hospitalization, prolonged jaundice, and respiratory distress. Neurological examination was normal. On MRI, T2-weighted (T2W) coronal images showed bilateral hyperintense cysts located superolaterally to frontal horns ( ► Fig. 1A and B ). Similarly axialT2Wimagesshowedthehyperintensecystanteriortothe frontal horns ( ► Fig. 1C ). On fl uid-attenuated inversion recovery coronal images and T1W, they appeared hypointense ( ► Fig. 2A and B ) without any evidence of associated in fl am-mation and periventricular gliosis. Given the location and no signi fi cantneonatalhistoryinthepatients,the fi ndingsonMRI are consistent with frontal horn cyst aka coarctation of lateral ventricles or connatal cysts. Connatal","PeriodicalId":51597,"journal":{"name":"Indian Journal of Radiology and Imaging","volume":"2020 375","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"136317038","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}
Mohsin Hussein, Stuart Metcalfe, Karthikeyan P. Iyengar, Rajesh Botchu
Abstract Tarsal coalition occurs in 1% of the population and represents a congenital failure of segmentation in two or more tarsal bones. It most commonly occurs at the talocalcaneal and calcaneonavicular joint. Although commonly asymptomatic, it may present with pain, rigidity, and pes planus. Cubonavicular, multiple synchronous, and bilateral coalitions are rare but an awareness is required to ensure accurate diagnosis and management. In this article, we presented the first reported case (to the best of our knowledge) of bilateral cubonavicular coalition with synchronous talocalcaneal coalition and stress response within the intermediate cuneiform.
{"title":"Bilateral Cubonavicular and Synchronous Talocalcaneal Tarsal Coalition with Stress Response—Case Report and Review of Literature","authors":"Mohsin Hussein, Stuart Metcalfe, Karthikeyan P. Iyengar, Rajesh Botchu","doi":"10.1055/s-0043-1776064","DOIUrl":"https://doi.org/10.1055/s-0043-1776064","url":null,"abstract":"Abstract Tarsal coalition occurs in 1% of the population and represents a congenital failure of segmentation in two or more tarsal bones. It most commonly occurs at the talocalcaneal and calcaneonavicular joint. Although commonly asymptomatic, it may present with pain, rigidity, and pes planus. Cubonavicular, multiple synchronous, and bilateral coalitions are rare but an awareness is required to ensure accurate diagnosis and management. In this article, we presented the first reported case (to the best of our knowledge) of bilateral cubonavicular coalition with synchronous talocalcaneal coalition and stress response within the intermediate cuneiform.","PeriodicalId":51597,"journal":{"name":"Indian Journal of Radiology and Imaging","volume":"54 8","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"136234972","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}
Abstract Partial anomalous pulmonary venous drainage is a congenital cardiac disorder characterized by abnormal drainage of one or more pulmonary veins into the systemic circulation. It can be isolated or associated with other congenital cardiac anomalies, most commonly atrial septal defect and patent ductus arteriosus. The clinical presentation is variable and depends on the degree of shunting and associated cardiac anomalies. Many patients usually remain asymptomatic until late in life. In this article, we presented a complex case of obstructive partial anomalous pulmonary venous drainage with dual drainage of bilateral accessory pulmonary veins with intact interatrial septum in conjunction with a patent ductus arteriosus and a ventricular septal defect. This pattern is incredibly rare and to the best of our knowledge has not been previously reported. Computed tomography played a pivotal role in precisely elucidating the intricate anatomy in this case with a complex pattern of anomalous pulmonary venous drainage.
{"title":"Hitherto Unreported Pattern of Complex Obstructive Partial Anomalous Pulmonary Venous Drainage with Dual Drainage of Accessory Pulmonary Veins","authors":"Dollphy Garg, Arun Sharma, Sanjeev Hanumantacharya Naganur, Manphool Singhal","doi":"10.1055/s-0043-1776107","DOIUrl":"https://doi.org/10.1055/s-0043-1776107","url":null,"abstract":"Abstract Partial anomalous pulmonary venous drainage is a congenital cardiac disorder characterized by abnormal drainage of one or more pulmonary veins into the systemic circulation. It can be isolated or associated with other congenital cardiac anomalies, most commonly atrial septal defect and patent ductus arteriosus. The clinical presentation is variable and depends on the degree of shunting and associated cardiac anomalies. Many patients usually remain asymptomatic until late in life. In this article, we presented a complex case of obstructive partial anomalous pulmonary venous drainage with dual drainage of bilateral accessory pulmonary veins with intact interatrial septum in conjunction with a patent ductus arteriosus and a ventricular septal defect. This pattern is incredibly rare and to the best of our knowledge has not been previously reported. Computed tomography played a pivotal role in precisely elucidating the intricate anatomy in this case with a complex pattern of anomalous pulmonary venous drainage.","PeriodicalId":51597,"journal":{"name":"Indian Journal of Radiology and Imaging","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"136316805","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}
Manasa C. Murthy, Bidisha Banerjee, Ullas Acharya, Shivakumar Shamarao
Abstract Acute encephalopathy with biphasic seizures and late reduced diffusion (AESD) is a clinicoradiological syndrome first recognized during the influenza pandemic in Japanese population in the late twentieth century.1 In this article, we presented a rare case report of AESD in a young child due to severe acute respiratory syndrome coronavirus 2 infection (SARS-CoV-2) who presented with febrile status epilepticus, persistent encephalopathy, and had recurrence of seizures on day 4 of illness with characteristic magnetic resonance imaging findings and a relatively fair outcome
{"title":"Acute Encephalopathy with Biphasic Seizures and Late Reduced Diffusion following SARS-CoV-2 Infection—A Rare Case Report","authors":"Manasa C. Murthy, Bidisha Banerjee, Ullas Acharya, Shivakumar Shamarao","doi":"10.1055/s-0043-1775797","DOIUrl":"https://doi.org/10.1055/s-0043-1775797","url":null,"abstract":"Abstract Acute encephalopathy with biphasic seizures and late reduced diffusion (AESD) is a clinicoradiological syndrome first recognized during the influenza pandemic in Japanese population in the late twentieth century.1 In this article, we presented a rare case report of AESD in a young child due to severe acute respiratory syndrome coronavirus 2 infection (SARS-CoV-2) who presented with febrile status epilepticus, persistent encephalopathy, and had recurrence of seizures on day 4 of illness with characteristic magnetic resonance imaging findings and a relatively fair outcome","PeriodicalId":51597,"journal":{"name":"Indian Journal of Radiology and Imaging","volume":"9 6","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"136234407","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}
Congenital cardiovascular lesions can be complex and perplexing to diagnose even with multiple imaging and post-processing modalities. Herein, we present a case of persistent fifth aortic arch (PFAA) with coarctation of aorta emphasizing role of computed tomography (CT) cinematic rendering to simplify the complex anatomy
{"title":"Cinematic Rendering of Persistent Fifth Aortic Arch with Aortic Coarctation and Bicuspid Aortic Valve","authors":"Karamkar Deepmala, Jineesh Valakkada, Anoop Ayappan, Basavaraj Birardar","doi":"10.1055/s-0043-1775798","DOIUrl":"https://doi.org/10.1055/s-0043-1775798","url":null,"abstract":"Congenital cardiovascular lesions can be complex and perplexing to diagnose even with multiple imaging and post-processing modalities. Herein, we present a case of persistent fifth aortic arch (PFAA) with coarctation of aorta emphasizing role of computed tomography (CT) cinematic rendering to simplify the complex anatomy","PeriodicalId":51597,"journal":{"name":"Indian Journal of Radiology and Imaging","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135853603","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}
Abstract Background Many different risk stratification systems have been formulated for thyroid nodules, differing in their fine-needle aspiration cytology (FNAC) indication, suggesting a lack of consensus around the world. Purpose This prospective study was conducted to find the best guideline for risk stratification, for a better malignancy yield, and with reduced rates of negative FNACs among three Thyroid Imaging, Reporting, and Data System (TIRADS) guidelines. Materials and Methods A total of 625 thyroid nodules with conclusive FNAC or histopathological diagnosis were included in the study. Various sonographic parameters were recorded. They were classified into categories as per the three guidelines and compared with FNAC diagnosis. The guidelines were evaluated in terms of sensitivity, specificity, predictive values, and diagnostic accuracy. Sensitivity and specificity were compared by McNemar's test. Results American College of Radiology (ACR) TIRADS had the highest diagnostic accuracy (56.8%), specificity (50.75%), positive predictive value (23.92%), lowest rates of negative FNACs (76.08%), and high negative predictive value (97.84 %). Korean (K) TIRADS had the maximum sensitivity (97.75%), highest negative predictive value (98.44%), and gross malignancy yield. European TIRADS was between the two other guidelines in most parameters with specificity like K TIRADS. Conclusion All the three guidelines are very good screening tools, with comparable high sensitivity. ACR TIRADS is better in terms of specificity and reduced rates of negative FNACs. Including the presence of a suspicious cervical lymph node as a criterion and more frequent follow-up might further improve the diagnostic performance of the guideline.
背景:针对甲状腺结节制定了许多不同的风险分层系统，其细针穿刺细胞学(FNAC)适应症不同，表明在世界范围内缺乏共识。目的:本前瞻性研究旨在寻找三种甲状腺成像、报告和数据系统(TIRADS)指南中风险分层的最佳指南，以获得更好的恶性肿瘤发生率，并降低FNACs阴性率。材料与方法本研究共纳入625例经FNAC或组织病理学确诊的甲状腺结节。记录各种超声参数。将其按照三个指南进行分类，并与FNAC诊断进行比较。根据敏感性、特异性、预测值和诊断准确性对指南进行评估。采用McNemar试验比较敏感性和特异性。结果美国放射学会(American College of Radiology, ACR) TIRADS诊断准确率最高(56.8%)，特异性最高(50.75%)，阳性预测值最高(23.92%)，FNACs阴性率最低(76.08%)，阴性预测值最高(97.84%)。韩国(K) TIRADS具有最高的敏感性(97.75%)，最高的阴性预测值(98.44%)和总恶性肿瘤发生率。欧洲TIRADS与K TIRADS一样，在大多数具有特异性的参数上介于其他两种指南之间。结论3种指南都是很好的筛查工具，具有相当高的灵敏度。ACR TIRADS在特异性和降低FNACs阴性率方面更好。将可疑颈淋巴结的存在作为一项标准和更频繁的随访可能会进一步提高指南的诊断效果。
{"title":"Determining the Best Thyroid Imaging Reporting and Data System: A Prospective Study Comparing the Diagnostic Performance of ACR, EU, and K TIRADS in the Evaluation of Thyroid Nodules","authors":"Supraja Laguduva Mohan, Ramkumar Govindarajalou, Dukhabandhu Naik, Sunil Kumar Saxena, Pampa Ch. Toi, Gomathi Shankar V.","doi":"10.1055/s-0043-1775862","DOIUrl":"https://doi.org/10.1055/s-0043-1775862","url":null,"abstract":"Abstract Background Many different risk stratification systems have been formulated for thyroid nodules, differing in their fine-needle aspiration cytology (FNAC) indication, suggesting a lack of consensus around the world. Purpose This prospective study was conducted to find the best guideline for risk stratification, for a better malignancy yield, and with reduced rates of negative FNACs among three Thyroid Imaging, Reporting, and Data System (TIRADS) guidelines. Materials and Methods A total of 625 thyroid nodules with conclusive FNAC or histopathological diagnosis were included in the study. Various sonographic parameters were recorded. They were classified into categories as per the three guidelines and compared with FNAC diagnosis. The guidelines were evaluated in terms of sensitivity, specificity, predictive values, and diagnostic accuracy. Sensitivity and specificity were compared by McNemar's test. Results American College of Radiology (ACR) TIRADS had the highest diagnostic accuracy (56.8%), specificity (50.75%), positive predictive value (23.92%), lowest rates of negative FNACs (76.08%), and high negative predictive value (97.84 %). Korean (K) TIRADS had the maximum sensitivity (97.75%), highest negative predictive value (98.44%), and gross malignancy yield. European TIRADS was between the two other guidelines in most parameters with specificity like K TIRADS. Conclusion All the three guidelines are very good screening tools, with comparable high sensitivity. ACR TIRADS is better in terms of specificity and reduced rates of negative FNACs. Including the presence of a suspicious cervical lymph node as a criterion and more frequent follow-up might further improve the diagnostic performance of the guideline.","PeriodicalId":51597,"journal":{"name":"Indian Journal of Radiology and Imaging","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135855263","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}
Abstract Background Although abundant literature is currently available on the use of deep learning for breast cancer detection in mammography, the quality of such literature is widely variable. Purpose To evaluate published literature on breast cancer detection in mammography for reproducibility and to ascertain best practices for model design. Methods The PubMed and Scopus databases were searched to identify records that described the use of deep learning to detect lesions or classify images into cancer or noncancer. A modification of Quality Assessment of Diagnostic Accuracy Studies (mQUADAS-2) tool was developed for this review and was applied to the included studies. Results of reported studies (area under curve [AUC] of receiver operator curve [ROC] curve, sensitivity, specificity) were recorded. Results A total of 12,123 records were screened, of which 107 fit the inclusion criteria. Training and test datasets, key idea behind model architecture, and results were recorded for these studies. Based on mQUADAS-2 assessment, 103 studies had high risk of bias due to nonrepresentative patient selection. Four studies were of adequate quality, of which three trained their own model, and one used a commercial network. Ensemble models were used in two of these. Common strategies used for model training included patch classifiers, image classification networks (ResNet in 67%), and object detection networks (RetinaNet in 67%). The highest reported AUC was 0.927 ± 0.008 on a screening dataset, while it reached 0.945 (0.919–0.968) on an enriched subset. Higher values of AUC (0.955) and specificity (98.5%) were reached when combined radiologist and Artificial Intelligence readings were used than either of them alone. None of the studies provided explainability beyond localization accuracy. None of the studies have studied interaction between AI and radiologist in a real world setting. Conclusion While deep learning holds much promise in mammography interpretation, evaluation in a reproducible clinical setting and explainable networks are the need of the hour.
{"title":"Reproducibility and Explainability of Deep Learning in Mammography: A Systematic Review of Literature","authors":"Deeksha Bhalla, Krithika Rangarajan, Tany Chandra, Subhashis Banerjee, Chetan Arora","doi":"10.1055/s-0043-1775737","DOIUrl":"https://doi.org/10.1055/s-0043-1775737","url":null,"abstract":"Abstract Background Although abundant literature is currently available on the use of deep learning for breast cancer detection in mammography, the quality of such literature is widely variable. Purpose To evaluate published literature on breast cancer detection in mammography for reproducibility and to ascertain best practices for model design. Methods The PubMed and Scopus databases were searched to identify records that described the use of deep learning to detect lesions or classify images into cancer or noncancer. A modification of Quality Assessment of Diagnostic Accuracy Studies (mQUADAS-2) tool was developed for this review and was applied to the included studies. Results of reported studies (area under curve [AUC] of receiver operator curve [ROC] curve, sensitivity, specificity) were recorded. Results A total of 12,123 records were screened, of which 107 fit the inclusion criteria. Training and test datasets, key idea behind model architecture, and results were recorded for these studies. Based on mQUADAS-2 assessment, 103 studies had high risk of bias due to nonrepresentative patient selection. Four studies were of adequate quality, of which three trained their own model, and one used a commercial network. Ensemble models were used in two of these. Common strategies used for model training included patch classifiers, image classification networks (ResNet in 67%), and object detection networks (RetinaNet in 67%). The highest reported AUC was 0.927 ± 0.008 on a screening dataset, while it reached 0.945 (0.919–0.968) on an enriched subset. Higher values of AUC (0.955) and specificity (98.5%) were reached when combined radiologist and Artificial Intelligence readings were used than either of them alone. None of the studies provided explainability beyond localization accuracy. None of the studies have studied interaction between AI and radiologist in a real world setting. Conclusion While deep learning holds much promise in mammography interpretation, evaluation in a reproducible clinical setting and explainable networks are the need of the hour.","PeriodicalId":51597,"journal":{"name":"Indian Journal of Radiology and Imaging","volume":"84 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"136294069","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：