Comparison of Volumetric and 2D Measurements and Longitudinal Trajectories in the Response Assessment of BRAF V600E-Mutant Pediatric Gliomas in the Pacific Pediatric Neuro-Oncology Consortium Clinical Trial.

BACKGROUND AND PURPOSE: Response on imaging is widely used to evaluate treatment efficacy in clinical trials of pediatric gliomas. While conventional criteria rely on 2D measurements, volumetric analysis may provide a more comprehensive response assessment. There is sparse research on the role of volumetrics in pediatric gliomas. Our purpose was to compare 2D and volumetric analysis with the assessment of neuroradiologists using the Brain Tumor Reporting and Data System (BT-RADS) in BRAF V600E-mutant pediatric gliomas. MATERIALS AND METHODS: Manual volumetric segmentations of whole and solid tumors were compared with 2D measurements in 31 participants (292 follow-up studies) in the Pacific Pediatric Neuro-Oncology Consortium 002 trial (NCT01748149). Two neuroradiologists evaluated responses using BT-RADS. Receiver operating characteristic analysis compared classification performance of 2D and volumetrics for partial response. Agreement between volumetric and 2D mathematically modeled longitudinal trajectories for 25 participants was determined using the model-estimated time to best response. RESULTS: Of 31 participants, 20 had partial responses according to BT-RADS criteria. Receiver operating characteristic curves for the classification of partial responders at the time of first detection (median = 2 months) yielded an area under the curve of 0.84 (95% CI, 0.69-0.99) for 2D area, 0.91 (95% CI, 0.80-1.00) for whole-volume, and 0.92 (95% CI, 0.82-1.00) for solid volume change. There was no significant difference in the area under the curve between 2D and solid (P = .34) or whole volume (P = .39). There was no significant correlation in model-estimated time to best response (ρ = 0.39, P >.05) between 2D and whole-volume trajectories. Eight of the 25 participants had a difference of ≥90 days in transition from partial response to stable disease between their 2D and whole-volume modeled trajectories. CONCLUSIONS: Although there was no overall difference between volumetrics and 2D in classifying partial response assessment using BT-RADS, further prospective studies will be critical to elucidate how the observed differences in tumor 2D and volumetric trajectories affect clinical decision-making and outcomes in some individuals.

A large open access dataset of brain metastasis 3D segmentations on MRI with clinical and imaging information.

Resection and whole brain radiotherapy (WBRT) are standard treatments for brain metastases (BM) but are associated with cognitive side effects. Stereotactic radiosurgery (SRS) uses a targeted approach with less side effects than WBRT. SRS requires precise identification and delineation of BM. While artificial intelligence (AI) algorithms have been developed for this, their clinical adoption is limited due to poor model performance in the clinical setting. The limitations of algorithms are often due to the quality of datasets used for training the AI network. The purpose of this study was to create a large, heterogenous, annotated BM dataset for training and validation of AI models. We present a BM dataset of 200 patients with pretreatment T1, T1 post-contrast, T2, and FLAIR MR images. The dataset includes contrast-enhancing and necrotic 3D segmentations on T1 post-contrast and peritumoral edema 3D segmentations on FLAIR. Our dataset contains 975 contrast-enhancing lesions, many of which are sub centimeter, along with clinical and imaging information. We used a streamlined approach to database-building through a PACS-integrated segmentation workflow.

Comparison of volumetric and 2D-based response methods in the PNOC-001 pediatric low-grade glioma clinical trial.

Background: Although response in pediatric low-grade glioma (pLGG) includes volumetric assessment, more simplified 2D-based methods are often used in clinical trials. The study's purpose was to compare volumetric to 2D methods. Methods: An expert neuroradiologist performed solid and whole tumor (including cyst and edema) volumetric measurements on MR images using a PACS-based manual segmentation tool in 43 pLGG participants (213 total follow-up images) from the Pacific Pediatric Neuro-Oncology Consortium (PNOC-001) trial. Classification based on changes in volumetric and 2D measurements of solid tumor were compared to neuroradiologist visual response assessment using the Brain Tumor Reporting and Data System (BT-RADS) criteria for a subset of 65 images using receiver operating characteristic (ROC) analysis. Longitudinal modeling of solid tumor volume was used to predict BT-RADS classification in 54 of the 65 images. Results: There was a significant difference in ROC area under the curve between 3D solid tumor volume and 2D area (0.96 vs 0.78, P = .005) and between 3D solid and 3D whole volume (0.96 vs 0.84, P = .006) when classifying BT-RADS progressive disease (PD). Thresholds of 15-25% increase in 3D solid tumor volume had an 80% sensitivity in classifying BT-RADS PD included in their 95% confidence intervals. The longitudinal model of solid volume response had a sensitivity of 82% and a positive predictive value of 67% for detecting BT-RADS PD. Conclusions: Volumetric analysis of solid tumor was significantly better than 2D measurements in classifying tumor progression as determined by BT-RADS criteria and will enable more comprehensive clinical management.

Application of novel PACS-based informatics platform to identify imaging based predictors of CDKN2A allelic status in glioblastomas.

Gliomas with CDKN2A mutations are known to have worse prognosis but imaging features of these gliomas are unknown. Our goal is to identify CDKN2A specific qualitative imaging biomarkers in glioblastomas using a new informatics workflow that enables rapid analysis of qualitative imaging features with Visually AcceSAble Rembrandtr Images (VASARI) for large datasets in PACS. Sixty nine patients undergoing GBM resection with CDKN2A status determined by whole-exome sequencing were included. GBMs on magnetic resonance images were automatically 3D segmented using deep learning algorithms incorporated within PACS. VASARI features were assessed using FHIR forms integrated within PACS. GBMs without CDKN2A alterations were significantly larger (64 vs. 30%, p = 0.007) compared to tumors with homozygous deletion (HOMDEL) and heterozygous loss (HETLOSS). Lesions larger than 8 cm were four times more likely to have no CDKN2A alteration (OR: 4.3; 95% CI 1.5-12.1; p < 0.001). We developed a novel integrated PACS informatics platform for the assessment of GBM molecular subtypes and show that tumors with HOMDEL are more likely to have radiographic evidence of pial invasion and less likely to have deep white matter invasion or subependymal invasion. These imaging features may allow noninvasive identification of CDKN2A allele status.

Integrating [18F]-Fluorodeoxyglucose Positron Emission Tomography with Computed Tomography with Radiation Therapy and Immunomodulation in Precision Therapy for Solid Tumors.

[18F]-FDG positron emission tomography with computed tomography (PET/CT) imaging is widely used to enhance the quality of care in patients diagnosed with cancer. Furthermore, it holds the potential to offer insight into the synergic effect of combining radiation therapy (RT) with immuno-oncological (IO) agents. This is achieved by evaluating treatment responses both at the RT and distant tumor sites, thereby encompassing the phenomenon known as the abscopal effect. In this context, PET/CT can play an important role in establishing timelines for RT/IO administration and monitoring responses, including novel patterns such as hyperprogression, oligoprogression, and pseudoprogression, as well as immune-related adverse events. In this commentary, we explore the incremental value of PET/CT to enhance the combination of RT with IO in precision therapy for solid tumors, by offering supplementary insights to recently released joint guidelines.

Evolution and implementation of radiographic response criteria in neuro-oncology.

Radiographic response assessment in neuro-oncology is critical in clinical practice and trials. Conventional criteria, such as the MacDonald and response assessment in neuro-oncology (RANO) criteria, rely on bidimensional (2D) measurements of a single tumor cross-section. Although RANO criteria are established for response assessment in clinical trials, there is a critical need to address the complexity of brain tumor treatment response with multiple new approaches being proposed. These include volumetric analysis of tumor compartments, structured MRI reporting systems like the Brain Tumor Reporting and Data System, and standardized approaches to advanced imaging techniques to distinguish tumor response from treatment effects. In this review, we discuss the strengths and limitations of different neuro-oncology response criteria and summarize current research findings on the role of novel response methods in neuro-oncology clinical trials and practice.

PACS-Integrated Tools for Peritumoral Edema Volumetrics Provide Additional Information to RANO-BM-Based Assessment of Lung Cancer Brain Metastases after Stereotactic Radiotherapy: A Pilot Study.

Stereotactic radiotherapy (SRT) is the standard of care treatment for brain metastases (METS) today. Nevertheless, there is limited understanding of how posttreatment lesional volumetric changes may assist prediction of lesional outcome. This is partly due to the paucity of volumetric segmentation tools. Edema alone can cause significant clinical symptoms and, therefore, needs independent study along with standard measurements of contrast-enhancing tumors. In this study, we aimed to compare volumetric changes of edema to RANO-BM-based measurements of contrast-enhancing lesion size. Patients with NSCLC METS ≥10 mm on post-contrast T1-weighted image and treated with SRT had measurements for up to seven follow-up scans using a PACS-integrated tool segmenting the peritumoral FLAIR hyperintense volume. Two-dimensional contrast-enhancing and volumetric edema changes were compared by creating treatment response curves. Fifty NSCLC METS were included in the study. The initial median peritumoral edema volume post-SRT relative to pre-SRT baseline was 37% (IQR 8-114%). Most of the lesions with edema volume reduction post-SRT experienced no increase in edema during the study. In over 50% of METS, the pattern of edema volume change was different than the pattern of contrast-enhancing lesion change at different timepoints, which was defined as incongruent. Lesions demonstrating incongruence at the first follow-up were more likely to progress subsequently. Therefore, edema assessment of METS post-SRT provides critical additional information to RANO-BM.

A Large Open Access Dataset of Brain Metastasis 3D Segmentations with Clinical and Imaging Feature Information.

Resection and whole brain radiotherapy (WBRT) are the standards of care for the treatment of patients with brain metastases (BM) but are often associated with cognitive side effects. Stereotactic radiosurgery (SRS) involves a more targeted treatment approach and has been shown to avoid the side effects associated with WBRT. However, SRS requires precise identification and delineation of BM. While many AI algorithms have been developed for this purpose, their clinical adoption has been limited due to poor model performance in the clinical setting. Major reasons for non-generalizable algorithms are the limitations in the datasets used for training the AI network. The purpose of this study was to create a large, heterogenous, annotated BM dataset for training and validation of AI models to improve generalizability. We present a BM dataset of 200 patients with pretreatment T1, T1 post-contrast, T2, and FLAIR MR images. The dataset includes contrast-enhancing and necrotic 3D segmentations on T1 post-contrast and whole tumor (including peritumoral edema) 3D segmentations on FLAIR. Our dataset contains 975 contrast-enhancing lesions, many of which are sub centimeter, along with clinical and imaging feature information. We used a streamlined approach to database-building leveraging a PACS-integrated segmentation workflow.

Systematic Literature Review of Machine Learning Algorithms Using Pretherapy Radiologic Imaging for Glioma Molecular Subtype Prediction.

BACKGROUND: The molecular profile of gliomas is a prognostic indicator for survival, driving clinical decision-making for treatment. Pathology-based molecular diagnosis is challenging because of the invasiveness of the procedure, exclusion from neoadjuvant therapy options, and the heterogeneous nature of the tumor. PURPOSE: We performed a systematic review of algorithms that predict molecular subtypes of gliomas from MR Imaging. DATA SOURCES: Data sources were Ovid Embase, Ovid MEDLINE, Cochrane Central Register of Controlled Trials, Web of Science. STUDY SELECTION: Per the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, 12,318 abstracts were screened and 1323 underwent full-text review, with 85 articles meeting the inclusion criteria. DATA ANALYSIS: We compared prediction results from different machine learning approaches for predicting molecular subtypes of gliomas. Bias analysis was conducted for each study, following the Prediction model Risk Of Bias Assessment Tool (PROBAST) guidelines. DATA SYNTHESIS: Isocitrate dehydrogenase mutation status was reported with an area under the curve and accuracy of 0.88 and 85% in internal validation and 0.86 and 87% in limited external validation data sets, respectively. For the prediction of O6-methylguanine-DNA methyltransferase promoter methylation, the area under the curve and accuracy in internal validation data sets were 0.79 and 77%, and in limited external validation, 0.89 and 83%, respectively. PROBAST scoring demonstrated high bias in all articles. LIMITATIONS: The low number of external validation and studies with incomplete data resulted in unequal data analysis. Comparing the best prediction pipelines of each study may introduce bias. CONCLUSIONS: While the high area under the curve and accuracy for the prediction of molecular subtypes of gliomas are reported in internal and external validation data sets, limited use of external validation and the increased risk of bias in all articles may present obstacles for clinical translation of these techniques.

Unlocking the Power of ChatGPT, Artificial Intelligence, and Large Language Models: Practical Suggestions for Radiation Oncologists.

Recent advances in artificial intelligence (AI), such as generative AI and large language models (LLMs), have generated significant excitement about the potential of AI to revolutionize our lives, work, and interaction with technology. This article explores the practical applications of LLMs, particularly ChatGPT, in the field of radiation oncology. We offer a guide on how radiation oncologists can interact with LLMs like ChatGPT in their routine clinical and administrative tasks, highlighting potential use cases of the present and future. We also highlight limitations and ethical considerations, including the current state of LLMs in decision making, protection of sensitive data, and the important role of human review of AI-generated content.

Clinical Informatics Approaches to Facilitate Cancer Data Sharing.

OBJECTIVES: Despite growing enthusiasm surrounding the utility of clinical informatics to improve cancer outcomes, data availability remains a persistent bottleneck to progress. Difficulty combining data with protected health information often limits our ability to aggregate larger more representative datasets for analysis. With the rise of machine learning techniques that require increasing amounts of clinical data, these barriers have magnified. Here, we review recent efforts within clinical informatics to address issues related to safely sharing cancer data. METHODS: We carried out a narrative review of clinical informatics studies related to sharing protected health data within cancer studies published from 2018-2022, with a focus on domains such as decentralized analytics, homomorphic encryption, and common data models. RESULTS: Clinical informatics studies that investigated cancer data sharing were identified. A particular focus of the search yielded studies on decentralized analytics, homomorphic encryption, and common data models. Decentralized analytics has been prototyped across genomic, imaging, and clinical data with the most advances in diagnostic image analysis. Homomorphic encryption was most often employed on genomic data and less on imaging and clinical data. Common data models primarily involve clinical data from the electronic health record. Although all methods have robust research, there are limited studies showing wide scale implementation. CONCLUSIONS: Decentralized analytics, homomorphic encryption, and common data models represent promising solutions to improve cancer data sharing. Promising results thus far have been limited to smaller settings. Future studies should be focused on evaluating the scalability and efficacy of these methods across clinical settings of varying resources and expertise.

Operational Ontology for Oncology (O3): A Professional Society-Based, Multistakeholder, Consensus-Driven Informatics Standard Supporting Clinical and Research Use of Real-World Data From Patients Treated for Cancer.

PURPOSE: The ongoing lack of data standardization severely undermines the potential for automated learning from the vast amount of information routinely archived in electronic health records (EHRs), radiation oncology information systems, treatment planning systems, and other cancer care and outcomes databases. We sought to create a standardized ontology for clinical data, social determinants of health, and other radiation oncology concepts and interrelationships. METHODS AND MATERIALS: The American Association of Physicists in Medicine's Big Data Science Committee was initiated in July 2019 to explore common ground from the stakeholders' collective experience of issues that typically compromise the formation of large inter- and intra-institutional databases from EHRs. The Big Data Science Committee adopted an iterative, cyclical approach to engaging stakeholders beyond its membership to optimize the integration of diverse perspectives from the community. RESULTS: We developed the Operational Ontology for Oncology (O3), which identified 42 key elements, 359 attributes, 144 value sets, and 155 relationships ranked in relative importance of clinical significance, likelihood of availability in EHRs, and the ability to modify routine clinical processes to permit aggregation. Recommendations are provided for best use and development of the O3 to 4 constituencies: device manufacturers, centers of clinical care, researchers, and professional societies. CONCLUSIONS: O3 is designed to extend and interoperate with existing global infrastructure and data science standards. The implementation of these recommendations will lower the barriers for aggregation of information that could be used to create large, representative, findable, accessible, interoperable, and reusable data sets to support the scientific objectives of grant programs. The construction of comprehensive "real-world" data sets and application of advanced analytical techniques, including artificial intelligence, holds the potential to revolutionize patient management and improve outcomes by leveraging increased access to information derived from larger, more representative data sets.

Screening for extranodal extension in HPV-associated oropharyngeal carcinoma: evaluation of a CT-based deep learning algorithm in patient data from a multicentre, randomised de-escalation trial.

BACKGROUND: Pretreatment identification of pathological extranodal extension (ENE) would guide therapy de-escalation strategies for in human papillomavirus (HPV)-associated oropharyngeal carcinoma but is diagnostically challenging. ECOG-ACRIN Cancer Research Group E3311 was a multicentre trial wherein patients with HPV-associated oropharyngeal carcinoma were treated surgically and assigned to a pathological risk-based adjuvant strategy of observation, radiation, or concurrent chemoradiation. Despite protocol exclusion of patients with overt radiographic ENE, more than 30% had pathological ENE and required postoperative chemoradiation. We aimed to evaluate a CT-based deep learning algorithm for prediction of ENE in E3311, a diagnostically challenging cohort wherein algorithm use would be impactful in guiding decision-making. METHODS: For this retrospective evaluation of deep learning algorithm performance, we obtained pretreatment CTs and corresponding surgical pathology reports from the multicentre, randomised de-escalation trial E3311. All enrolled patients on E3311 required pretreatment and diagnostic head and neck imaging; patients with radiographically overt ENE were excluded per study protocol. The lymph node with largest short-axis diameter and up to two additional nodes were segmented on each scan and annotated for ENE per pathology reports. Deep learning algorithm performance for ENE prediction was compared with four board-certified head and neck radiologists. The primary endpoint was the area under the curve (AUC) of the receiver operating characteristic. FINDINGS: From 178 collected scans, 313 nodes were annotated: 71 (23%) with ENE in general, 39 (13%) with ENE larger than 1 mm ENE. The deep learning algorithm AUC for ENE classification was 0·86 (95% CI 0·82-0·90), outperforming all readers (p<0·0001 for each). Among radiologists, there was high variability in specificity (43-86%) and sensitivity (45-96%) with poor inter-reader agreement (κ 0·32). Matching the algorithm specificity to that of the reader with highest AUC (R2, false positive rate 22%) yielded improved sensitivity to 75% (+ 13%). Setting the algorithm false positive rate to 30% yielded 90% sensitivity. The algorithm showed improved performance compared with radiologists for ENE larger than 1 mm (p<0·0001) and in nodes with short-axis diameter 1 cm or larger. INTERPRETATION: The deep learning algorithm outperformed experts in predicting pathological ENE on a challenging cohort of patients with HPV-associated oropharyngeal carcinoma from a randomised clinical trial. Deep learning algorithms should be evaluated prospectively as a treatment selection tool. FUNDING: ECOG-ACRIN Cancer Research Group and the National Cancer Institute of the US National Institutes of Health.

Using ChatGPT to evaluate cancer myths and misconceptions: artificial intelligence and cancer information.

Data about the quality of cancer information that chatbots and other artificial intelligence systems provide are limited. Here, we evaluate the accuracy of cancer information on ChatGPT compared with the National Cancer Institute's (NCI's) answers by using the questions on the "Common Cancer Myths and Misconceptions" web page. The NCI's answers and ChatGPT answers to each question were blinded, and then evaluated for accuracy (accurate: yes vs no). Ratings were evaluated independently for each question, and then compared between the blinded NCI and ChatGPT answers. Additionally, word count and Flesch-Kincaid readability grade level for each individual response were evaluated. Following expert review, the percentage of overall agreement for accuracy was 100% for NCI answers and 96.9% for ChatGPT outputs for questions 1 through 13 (ĸ = ‒0.03, standard error = 0.08). There were few noticeable differences in the number of words or the readability of the answers from NCI or ChatGPT. Overall, the results suggest that ChatGPT provides accurate information about common cancer myths and misconceptions.

Comparing Detection Schemes for Adversarial Images against Deep Learning Models for Cancer Imaging.

Deep learning (DL) models have demonstrated state-of-the-art performance in the classification of diagnostic imaging in oncology. However, DL models for medical images can be compromised by adversarial images, where pixel values of input images are manipulated to deceive the DL model. To address this limitation, our study investigates the detectability of adversarial images in oncology using multiple detection schemes. Experiments were conducted on thoracic computed tomography (CT) scans, mammography, and brain magnetic resonance imaging (MRI). For each dataset we trained a convolutional neural network to classify the presence or absence of malignancy. We trained five DL and machine learning (ML)-based detection models and tested their performance in detecting adversarial images. Adversarial images generated using projected gradient descent (PGD) with a perturbation size of 0.004 were detected by the ResNet detection model with an accuracy of 100% for CT, 100% for mammogram, and 90.0% for MRI. Overall, adversarial images were detected with high accuracy in settings where adversarial perturbation was above set thresholds. Adversarial detection should be considered alongside adversarial training as a defense technique to protect DL models for cancer imaging classification from the threat of adversarial images.

Comparing 3D, 2.5D, and 2D Approaches to Brain Image Auto-Segmentation.

Deep-learning methods for auto-segmenting brain images either segment one slice of the image (2D), five consecutive slices of the image (2.5D), or an entire volume of the image (3D). Whether one approach is superior for auto-segmenting brain images is not known. We compared these three approaches (3D, 2.5D, and 2D) across three auto-segmentation models (capsule networks, UNets, and nnUNets) to segment brain structures. We used 3430 brain MRIs, acquired in a multi-institutional study, to train and test our models. We used the following performance metrics: segmentation accuracy, performance with limited training data, required computational memory, and computational speed during training and deployment. The 3D, 2.5D, and 2D approaches respectively gave the highest to lowest Dice scores across all models. 3D models maintained higher Dice scores when the training set size was decreased from 3199 MRIs down to 60 MRIs. 3D models converged 20% to 40% faster during training and were 30% to 50% faster during deployment. However, 3D models require 20 times more computational memory compared to 2.5D or 2D models. This study showed that 3D models are more accurate, maintain better performance with limited training data, and are faster to train and deploy. However, 3D models require more computational memory compared to 2.5D or 2D models.

Developing Validated Tools to Identify Pulmonary Embolism in Electronic Databases: Rationale and Design of the PE-EHR+ Study.

BACKGROUND: Contemporary pulmonary embolism (PE) research, in many cases, relies on data from electronic health records (EHRs) and administrative databases that use International Classification of Diseases (ICD) codes. Natural language processing (NLP) tools can be used for automated chart review and patient identification. However, there remains uncertainty with the validity of ICD-10 codes or NLP algorithms for patient identification. METHODS: The PE-EHR+ study has been designed to validate ICD-10 codes as Principal Discharge Diagnosis, or Secondary Discharge Diagnoses, as well as NLP tools set out in prior studies to identify patients with PE within EHRs. Manual chart review by two independent abstractors by predefined criteria will be the reference standard. Sensitivity, specificity, and positive and negative predictive values will be determined. We will assess the discriminatory function of code subgroups for intermediate- and high-risk PE. In addition, accuracy of NLP algorithms to identify PE from radiology reports will be assessed. RESULTS: A total of 1,734 patients from the Mass General Brigham health system have been identified. These include 578 with ICD-10 Principal Discharge Diagnosis codes for PE, 578 with codes in the secondary position, and 578 without PE codes during the index hospitalization. Patients within each group were selected randomly from the entire pool of patients at the Mass General Brigham health system. A smaller subset of patients will also be identified from the Yale-New Haven Health System. Data validation and analyses will be forthcoming. CONCLUSIONS: The PE-EHR+ study will help validate efficient tools for identification of patients with PE in EHRs, improving the reliability of efficient observational studies or randomized trials of patients with PE using electronic databases.

Bladder Cancer Radiation Oncology of the Future: Prognostic Modelling, Radiomics, and Treatment Planning With Artificial Intelligence.

Machine learning (ML) and artificial intelligence (AI) have demonstrated potential to improve the care of radiation oncology patients. Here we review recent advances applicable to the care of bladder cancer, with an eye towards studies that may suggest next steps in clinical implementation. Algorithms have been applied to clinical records, pathology, and radiology data to generate accurate predictive models for prognosis and clinical outcomes. AI has also shown increasing utility for auto-contouring and efficient creation of workflows involving multiple treatment plans. As technologies progress towards routine clinical use for bladder cancer patients, we also discuss emerging methods to improve interpretability and reliability of algorithms.

Artificial Intelligence in Breast Cancer Screening: Evaluation of FDA Device Regulation and Future Recommendations.

Importance: Contemporary approaches to artificial intelligence (AI) based on deep learning have generated interest in the application of AI to breast cancer screening (BCS). The US Food and Drug Administration (FDA) has approved several next-generation AI products indicated for BCS in recent years; however, questions regarding their accuracy, appropriate use, and clinical utility remain. Objectives: To describe the current FDA regulatory process for AI products, summarize the evidence used to support FDA clearance and approval of AI products indicated for BCS, consider the advantages and limitations of current regulatory approaches, and suggest ways to improve the current system. Evidence Review: Premarket notifications and other publicly available documents used for FDA clearance and approval of AI products indicated for BCS from January 1, 2017, to December 31, 2021. Findings: Nine AI products indicated for BCS for identification of suggestive lesions and mammogram triage were included. Most of the products had been cleared through the 510(k) pathway, and all clearances were based on previously collected retrospective data; 6 products used multicenter designs; 7 products used enriched data; and 4 lacked details on whether products were externally validated. Test performance measures, including sensitivity, specificity, and area under the curve, were the main outcomes reported. Most of the devices used tissue biopsy as the criterion standard for BCS accuracy evaluation. Other clinical outcome measures, including cancer stage at diagnosis and interval cancer detection, were not reported for any of the devices. Conclusions and Relevance: The findings of this review suggest important gaps in reporting of data sources, data set type, validation approach, and clinical utility assessment. As AI-assisted reading becomes more widespread in BCS and other radiologic examinations, strengthened FDA evidentiary regulatory standards, development of postmarketing surveillance, a focus on clinically meaningful outcomes, and stakeholder engagement will be critical for ensuring the safety and efficacy of these products.