Purpose: To develop quantitative molecular imaging biomarkers of immune-related adverse event (irAE) development in malignant melanoma (MM) patients receiving immune-checkpoint inhibitors (ICI) imaged with ¹⁸F-FDG PET/CT. Methods: ¹⁸F-FDG PET/CT images of 58 MM patients treated with anti-PD-1 or anti-CTLA-4 ICI were retrospectively analyzed for indication of irAE. Three target organs, most commonly affected by irAE, were considered: bowel, lung, and thyroid. Patient charts were reviewed to identify which patients experienced irAE, irAE grade, and time to irAE diagnosis. Target organs were segmented using a convolutional neural network (CNN), and novel quantitative imaging biomarkers — SUV percentiles (SUV_X%) of ¹⁸F-FDG uptake within the target organs — were correlated with the clinical irAE status. Area under the receiver-operating characteristic curve (AUROC) was used to quantify irAE detection performance. Patients who did not experience irAE were used to establish normal ranges for target organ ¹⁸F-FDG uptake. Results: A total of 31% (18/58) patients experienced irAE in the three target organs: bowel (n=6), lung (n=5), and thyroid (n=9). Optimal percentiles for identifying irAE were bowel (SUV_95%, AUROC=0.79), lung (SUV_95%, AUROC=0.98), and thyroid (SUV_75%, AUROC=0.88). Optimal cut-offs for irAE detection were bowel (SUV_95%>2.7 g/mL), lung (SUV_95%>1.7 g/mL), and thyroid (SUV_75%>2.1 g/mL). Normal ranges (95% confidence interval) for the SUV percentiles in patients without irAE were bowel [1.74, 2.86 g/mL], lung [0.73, 1.46 g/mL], and thyroid [0.86, 1.99 g/mL]. Conclusions: Increased ¹⁸F-FDG uptake within irAE-affected organs provides predictive information about the development of irAE in MM patients receiving ICI and represents a potential quantitative imaging biomarker for irAE. Some irAE can be detected on ¹⁸F-FDG PET/CT well before clinical symptoms appear.

Objective . Deep Learning models are often susceptible to failures after deployment. Knowing when your model is producing inadequate predictions is crucial. In this work, we investigate the utility of Monte Carlo (MC) dropout and the efficacy of the proposed uncertainty metric (UM) for flagging of unacceptable pectoral muscle segmentations in mammograms. Approach . Segmentation of pectoral muscle was performed with modified ResNet18 convolutional neural network. MC dropout layers were kept unlocked at inference time. For each mammogram, 50 pectoral muscle segmentations were generated. The mean was used to produce the final segmentation and the standard deviation was applied for the estimation of uncertainty. From each pectoral muscle uncertainty map, the overall UM was calculated. To validate the UM, a correlation between the dice similarity coefficient (DSC) and UM was used. The UM was first validated in a training set (200 mammograms) and finally tested in an independent dataset (300 mammograms). ROC-AUC analysis was performed to test the discriminatory power of the proposed UM for flagging unacceptable segmentations. Main results . The introduction of dropout layers in the model improved segmentation performance (DSC = 0.95 ± 0.07 versus DSC = 0.93 ± 0.10). Strong anti-correlation ( r = −0.76, p < 0.001) between the proposed UM and DSC was observed. A high AUC of 0.98 (97% specificity at 100% sensitivity) was obtained for the discrimination of unacceptable segmentations. Qualitative inspection by the radiologist revealed that images with high UM are difficult to segment. Significance . The use of MC dropout at inference time in combination with the proposed UM enables flagging of unacceptable pectoral muscle segmentations from mammograms with excellent discriminatory power.

: Objective. Deep-learning-based models have achieved state-of-the-art breast cancer risk (BCR) prediction performance. However, these models are highly complex, and the underlying mechanisms of BCR prediction are not fully understood. Key questions include whether these models can detect breast morphologic changes that lead to cancer. These findings would boost confidence in utilizing BCR models in practice and provide clinicians with new perspectives. In this work, we aimed to determine when oncogenic processes in the breast provide sufficient signal for the models to detect these changes. Approach. In total, 1210 screening mammograms were collected for patients screened at different times before the cancer was screen-detected and 2400 mammograms for patients with at least ten years of follow-up. MIRAI, a BCR risk prediction model, was used to estimate the BCR. Attribution heterogeneity was defined as the relative difference between the attributions obtained from the right and left breasts using one of the eight interpretability techniques. Model reliance on the side of the breast with cancer was quantified with AUC. The Mann–Whitney U test was used to check for significant differences in median absolute Attribution Heterogeneity between cancer patients and healthy individuals. Results. All tested attribution methods showed a similar longitudinal trend, where the model reliance on the side of the breast with cancer was the highest for the 0–1 years-to-cancer interval (AUC = 0.85–0.95), dropped for the 1–3 years-to-cancer interval (AUC = 0.64–0.71), and remained above the threshold for random performance for the 3–5 years-to-cancer interval (AUC = 0.51–0.58). For all eight attribution methods, the median values of absolute attribution heterogeneity were significantly larger for patients diagnosed with cancer at one point (p < 0.01). Significance. Interpretability of BCR prediction has revealed that long-term predictions (beyond three years) are most likely based on typical breast characteristics, such as breast density; for mid-term predictions (one to three years), the model appears to detect early signs of tumor development, while for short-term predictions (up to a year), the BCR model essentially functions as a breast cancer detection model.

A feature-centric explainable artificial intelligence pipeline showed that the Mirai model uses lesion features on mammograms for lesion detection and breast cancer risk prediction. A simplified model using only calcification features achieved comparable risk prediction performance to the original Mirai model in a screen-negative population.

Objective. Deep learning models that aid in medical image assessment tasks must be both accurate and reliable to be deployed within clinical settings. While deep learning models have been shown to be highly accurate across a variety of tasks, measures that indicate the reliability of these models are less established. Increasingly, uncertainty quantification (UQ) methods are being introduced to inform users on the reliability of model outputs. However, most existing methods cannot be augmented to previously validated models because they are not post hoc, and they change a model’s output. In this work, we overcome these limitations by introducing a novel post hoc UQ method, termed Local Gradients UQ, and demonstrate its utility for deep learning-based metastatic disease delineation. Approach. This method leverages a trained model’s localized gradient space to assess sensitivities to trained model parameters. We compared the Local Gradients UQ method to non-gradient measures defined using model probability outputs. The performance of each uncertainty measure was assessed in four clinically relevant experiments: (1) response to artificially degraded image quality, (2) comparison between matched high- and low-quality clinical images, (3) false positive (FP) filtering, and (4) correspondence with physician-rated disease likelihood. Main results. (1) Response to artificially degraded image quality was enhanced by the Local Gradients UQ method, where the median percent difference between matching lesions in non-degraded and most degraded images was consistently higher for the Local Gradients uncertainty measure than the non-gradient uncertainty measures (e.g. 62.35% vs. 2.16% for additive Gaussian noise). (2) The Local Gradients UQ measure responded better to high- and low-quality clinical images (p < 0.05 vs p > 0.1 for both non-gradient uncertainty measures). (3) FP filtering performance was enhanced by the Local Gradients UQ method when compared to the non-gradient methods, increasing the area under the receiver operating characteristic curve (ROC AUC) by 20.1% and decreasing the false positive rate by 26%. (4) The Local Gradients UQ method also showed more favorable correspondence with physician-rated likelihood for malignant lesions by increasing ROC AUC for correspondence with physician-rated disease likelihood by 16.2%. Significance. In summary, this work introduces and validates a novel gradient-based UQ method for deep learning-based medical image assessments to enhance user trust when using deployed clinical models.

In this study, a comprehensive and realistic model of laser light interaction with skin and hair was constructed. The model was applied to study the characteristics of laser-tissue interaction for the deeply penetrating Nd:YAG laser. Three types of finite element method (FEM) models were developed. In the first model, the hair shaft grew straight out of the follicle; in the second model, it grew at a variable angle; and in the third model, an array of hair was considered. The transport equation and heat diffusion equation were solved with the mesh-based Monte Carlo method and partial differential equations, respectively. The results of the simulations indicated that the area of necrosis increased with increasing fluence; cooling had a limited effect on the extent of necrosis, particularly at a fluence of 80 J/cm². The thermal damage to hair follicles on the periphery of an irradiated array of hair may be insufficient for achieving necrosis. The pulse itself and the short cooling-down period after the pulse contributed the most to the final thermal damage to the hair follicle. The FEM modeling of laser-tissue interaction has proven to be a useful tool for studying the influence of different therapeutic parameters on the resulting hair and skin damage.