Cancer care increasingly relies on imaging for patient management. The two most common cross-sectional imaging modalities in oncology are computed tomography (CT) and magnetic resonance imaging (MRI), which provide high-resolution anatomic and physiological imaging. Herewith is a summary of recent applications of rapidly advancing artificial intelligence (AI) in CT and MRI oncological imaging that addresses the benefits and challenges of the resultant opportunities with examples. Major challenges remain, such as how best to integrate AI developments into clinical radiology practice, the vigorous assessment of quantitative CT and MR imaging data accuracy, and reliability for clinical utility and research integrity in oncology. Such challenges necessitate an evaluation of the robustness of imaging biomarkers to be included in AI developments, a culture of data sharing, and the cooperation of knowledgeable academics with vendor scientists and companies operating in radiology and oncology fields. Herein, we will illustrate a few challenges and solutions of these efforts using novel methods for synthesizing different contrast modality images, auto-segmentation, and image reconstruction with examples from lung CT as well as abdome, pelvis, and head and neck MRI. The imaging community must embrace the need for quantitative CT and MRI metrics beyond lesion size measurement. AI methods for the extraction and longitudinal tracking of imaging metrics from registered lesions and understanding the tumor environment will be invaluable for interpreting disease status and treatment efficacy. This is an exciting time to work together to move the imaging field forward with narrow AI-specific tasks. New AI developments using CT and MRI datasets will be used to improve the personalized management of cancer patients.

Computed Tomography (CT) contributes up to 50% of the medical exposure to the United States population. Children are considered to be at higher risk of developing radiation-induced tumors due to the young age of exposure and increased tissue radiosensitivity. Organ dose estimation is essential for pediatric and adult patient cancer risk assessment. The objective of this study is to validate the VirtualDose software in comparison to currently available software and methods for pediatric and adult CT organ dose estimation. Five age groups of pediatric patients and adult patients were simulated by three organ dose estimators. Head, chest, abdomen-pelvis, and chest-abdomen-pelvis CT scans were simulated, and doses to organs both inside and outside the scan range were compared. For adults, VirtualDose was compared against ImPACT and CT-Expo. For pediatric patients, VirtualDose was compared to CT-Expo and compared to size-based methods from literature. Pediatric to adult effective dose ratios were also calculated with VirtualDose, and were compared with the ranges of effective dose ratios provided in ImPACT. In-field organs see less than 60% difference in dose between dose estimators. For organs outside scan range or distributed organs, a five times’ difference can occur. VirtualDose agrees with the size-based methods within 20% difference for the organs investigated. Between VirtualDose and ImPACT, the pediatric to adult ratios for effective dose are compared, and less than 21% difference is observed for chest scan while more than 40% difference is observed for head-neck scan and abdomen-pelvis scan. For pediatric patients, 2 cm scan range change can lead to a five times dose difference in partially scanned organs. VirtualDose is validated against CT-Expo and ImPACT with relatively small discrepancies in dose for organs inside scan range, while large discrepancies in dose are observed for organs outside scan range. Patient-specific organ dose estimation is possible using the size-based methods, and VirtualDose agrees with size-based method for the organs investigated. Careful range selection for CT protocols is necessary for organ dose optimization for pediatric and adult patients.

Estimate organ and effective doses from computed tomography scans of pediatric oncologic patients using patient-specific information. With IRB approval patient-specific scan parameters and patient size obtained from DICOM images and vendor-provided dose monitoring application were obtained for a cross-sectional study of 1250 pediatric patients from 0 through 20 y-olds who underwent head, chest, abdomen-pelvis, or chest-abdomen-pelvis CT scans. Patients were categorized by age. Organ doses and effective doses were estimated using VirtualDose™ CT based on patient-specific information, tube current modulation (TCM), and age-specific realistic phantoms. CTDIvol, DLP, and dose results were compared with those reported in the literature. CTDIvol and DLP varied widely as patient size varied. The 75th percentiles of CTDIvol and DLP were no greater than in the literature with the exception of head scans of 16-20 y-olds and of abdomen-pelvis scans of larger patients. Eye lens dose from a head scan was up to 69 mGy. Mean organ doses agreed with other studies at maximal difference of 38% for chest and 41% for abdomen-pelvis scans. Mean effective dose was generally higher for older patients. The highest effective doses were estimated for the 16-20 y-olds as: head 3.3 mSv, chest 4.1 mSv, abdomen-pelvis 10.0 mSv, chest-abdomen-pelvis 14.0 mSv. Patient-specific organ and effective doses have been estimated for pediatric oncologic patients from <1 through 20 y-olds. The effect of TCM was successfully accounted for in the estimates. Output parameters varied with patient size. CTDIvol and DLP results are useful for future protocol optimization.

The adoption of artificial intelligence (AI) tools in medicine poses challenges to existing clinical workflows. This commentary discusses the necessity of context-specific quality assurance (QA), emphasizing the need for robust QA measures with quality control (QC) procedures that encompass (1) acceptance testing (AT) before clinical use, (2) continuous QC monitoring, and (3) adequate user training. The discussion also covers essential components of AT and QA, illustrated with real-world examples. We also highlight what we see as the shared responsibility of manufacturers or vendors, regulators, healthcare systems, medical physicists, and clinicians to enact appropriate testing and oversight to ensure a safe and equitable transformation of medicine through AI.

Innovation in medical imaging artificial intelligence (AI)/machine learning (ML) demands extensive data collection, algorithmic advancements, and rigorous performance assessments encompassing aspects such as generalizability, uncertainty, bias, fairness, trustworthiness, and interpretability. Achieving widespread integration of AI/ML algorithms into diverse clinical tasks will demand a steadfast commitment to overcoming issues in model design, development, and performance assessment. The complexities of AI/ML clinical translation present substantial challenges, requiring engagement with relevant stakeholders, assessment of cost-effectiveness for user and patient benefit, timely dissemination of information relevant to robust functioning throughout the AI/ML lifecycle, consideration of regulatory compliance, and feedback loops for real-world performance evidence. This commentary addresses several hurdles for the development and adoption of AI/ML technologies in medical imaging. Comprehensive attention to these underlying and often subtle factors is critical not only for tackling the challenges but also for exploring novel opportunities for the advancement of AI in radiology.

Artificial intelligence (AI) has been successful at solving numerous problems in machine perception. In radiology, AI systems are rapidly evolving and show progress in guiding treatment decisions, diagnosing, localizing disease on medical images, and improving radiologists' efficiency. A critical component to deploying AI in radiology is to gain confidence in a developed system's efficacy and safety. The current gold standard approach is to conduct an analytical validation of performance on a generalization dataset from one or more institutions, followed by a clinical validation study of the system's efficacy during deployment. Clinical validation studies are time-consuming, and best practices dictate limited re-use of analytical validation data, so it is ideal to know ahead of time if a system is likely to fail analytical or clinical validation. In this paper, we describe a series of sanity tests to identify when a system performs well on development data for the wrong reasons. We illustrate the sanity tests' value by designing a deep learning system to classify pancreatic cancer seen in computed tomography scans.