Dataset Information

Stress testing reveals gaps in clinic readiness of image-based diagnostic artificial intelligence models.

ABSTRACT: Artificial intelligence models match or exceed dermatologists in melanoma image classification. Less is known about their robustness against real-world variations, and clinicians may incorrectly assume that a model with an acceptable area under the receiver operating characteristic curve or related performance metric is ready for clinical use. Here, we systematically assessed the performance of dermatologist-level convolutional neural networks (CNNs) on real-world non-curated images by applying computational "stress tests". Our goal was to create a proxy environment in which to comprehensively test the generalizability of off-the-shelf CNNs developed without training or evaluation protocols specific to individual clinics. We found inconsistent predictions on images captured repeatedly in the same setting or subjected to simple transformations (e.g., rotation). Such transformations resulted in false positive or negative predictions for 6.5-22% of skin lesions across test datasets. Our findings indicate that models meeting conventionally reported metrics need further validation with computational stress tests to assess clinic readiness.

SUBMITTER: Young AT

PROVIDER: S-EPMC7820258 | biostudies-literature | 2021 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Stress testing reveals gaps in clinic readiness of image-based diagnostic artificial intelligence models.

Young Albert T AT Fernandez Kristen K Pfau Jacob J Reddy Rasika R Cao Nhat Anh NA von Franque Max Y MY Johal Arjun A Wu Benjamin V BV Wu Rachel R RR Chen Jennifer Y JY Fadadu Raj P RP Vasquez Juan A JA Tam Andrew A Keiser Michael J MJ Wei Maria L ML

NPJ digital medicine 20210121 1

Artificial intelligence models match or exceed dermatologists in melanoma image classification. Less is known about their robustness against real-world variations, and clinicians may incorrectly assume that a model with an acceptable area under the receiver operating characteristic curve or related performance metric is ready for clinical use. Here, we systematically assessed the performance of dermatologist-level convolutional neural networks (CNNs) on real-world non-curated images by applying ...[more]

PMID: 33479460

Similar Datasets

Project description:BACKGROUND. Potential approaches for abbreviated knee MRI, including prospective acceleration with deep learning, have achieved limited clinical implementation. OBJECTIVE. The objective of this study was to evaluate the interreader agreement between conventional knee MRI and a 5-minute 3D quantitative double-echo steady-state (qDESS) sequence with automatic T2 mapping and deep learning super-resolutionaugmentation and to compare the diagnostic performance of the two methods regarding findings from arthroscopic surgery. METHODS. Fifty-one patients with knee pain underwent knee MRI that included an additional 3D qDESS sequence with automatic T2 mapping. Fourier interpolation was followed by prospective deep learning super resolution to enhance qDESS slice resolution twofold. A musculoskeletal radiologist and a radiology resident performed independent retrospective evaluations of articular cartilage, menisci, ligaments, bones, extensor mechanism, and synovium using conventional MRI. Following a 2-month washout period, readers reviewed qDESS images alone followed by qDESS with the automatic T2 maps. Interreader agreement between conventional MRI and qDESS was computed using percentage agreement and Cohen kappa. The sensitivity and specificity of conventional MRI, qDESS alone, and qDESS plus T2 mapping were compared with arthroscopic findings using exact McNemar tests. RESULTS. Conventional MRI and qDESS showed 92% agreement in evaluating all tissues. Kappa was 0.79 (95% CI, 0.76-0.81) across all imaging findings. In 43 patients who underwent arthroscopy, sensitivity and specificity were not significantly different (p = .23 to > .99) between conventional MRI (sensitivity, 58-93%; specificity, 27-87%) and qDESS alone (sensitivity, 54-90%; specificity, 23-91%) for cartilage, menisci, ligaments, and synovium. For grade 1 cartilage lesions, sensitivity and specificity were 33% and 56%, respectively, for conventional MRI; 23% and 53% for qDESS (p = .81); and 46% and 39% for qDESS with T2 mapping (p = .80). For grade 2A lesions, values were 27% and 53% for conventional MRI, 26% and 52% for qDESS (p = .02), and 58% and 40% for qDESS with T2 mapping (p < .001). CONCLUSION. The qDESS method prospectively augmented with deep learning showed strong interreader agreement with conventional knee MRI and near-equivalent diagnostic performance regarding arthroscopy. The ability of qDESS to automatically generate T2 maps increases sensitivity for cartilage abnormalities. CLINICAL IMPACT. Using prospective artificial intelligence to enhance qDESS image quality may facilitate an abbreviated knee MRI protocol while generating quantitative T2 maps.

Project description:BackgroundSimilar to understanding how blood pressure is measured by a sphygmomanometer, physicians will soon have to understand how an artificial intelligence-based application has come to the conclusion that a patient has hypertension, diabetes, or cancer. Although there are an increasing number of use cases where artificial intelligence is or can be applied to improve medical outcomes, the extent to which medical doctors and students are ready to work and leverage this paradigm is unclear.ObjectiveThis research aims to capture medical students' and doctors' level of familiarity toward artificial intelligence in medicine as well as their challenges, barriers, and potential risks linked to the democratization of this new paradigm.MethodsA web-based questionnaire comprising five dimensions-demographics, concepts and definitions, training and education, implementation, and risks-was systematically designed from a literature search. It was completed by 207 participants in total, of which 105 (50.7%) medical doctors and 102 (49.3%) medical students trained in all continents, with most of them in Europe, the Middle East, Asia, and North America.ResultsThe results revealed no significant difference in the familiarity of artificial intelligence between medical doctors and students (P=.91), except that medical students perceived artificial intelligence in medicine to lead to higher risks for patients and the field of medicine in general (P<.001). We also identified a rather low level of familiarity with artificial intelligence (medical students=2.11/5; medical doctors=2.06/5) as well as a low attendance to education or training. Only 2.9% (3/105) of medical doctors attended a course on artificial intelligence within the previous year, compared with 9.8% (10/102) of medical students. The complexity of the field of medicine was considered one of the biggest challenges (medical doctors=3.5/5; medical students=3.8/5), whereas the reduction of physicians' skills was the most important risk (medical doctors=3.3; medical students=3.6; P=.03).ConclusionsThe question is not whether artificial intelligence will be used in medicine, but when it will become a standard practice for optimizing health care. The low level of familiarity with artificial intelligence identified in this study calls for the implementation of specific education and training in medical schools and hospitals to ensure that medical professionals can leverage this new paradigm and improve health outcomes.

Dataset Information

Stress testing reveals gaps in clinic readiness of image-based diagnostic artificial intelligence models.

Publications

Stress testing reveals gaps in clinic readiness of image-based diagnostic artificial intelligence models.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets