More Like this

1. Risk-Aware Machine Learning Classifier for Skin Lesion Diagnosis

   https://doi.org/10.3390/jcm8081241  

   Mobiny, Aryan; Singh, Aditi; Van Nguyen, Hien (August 2019, Journal of Clinical Medicine)

   null (Ed.)

   Knowing when a machine learning system is not confident about its prediction is crucial in medical domains where safety is critical. Ideally, a machine learning algorithm should make a prediction only when it is highly certain about its competency, and refer the case to physicians otherwise. In this paper, we investigate how Bayesian deep learning can improve the performance of the machine–physician team in the skin lesion classification task. We used the publicly available HAM10000 dataset, which includes samples from seven common skin lesion categories: Melanoma (MEL), Melanocytic Nevi (NV), Basal Cell Carcinoma (BCC), Actinic Keratoses and Intraepithelial Carcinoma (AKIEC), Benign Keratosis (BKL), Dermatofibroma (DF), and Vascular (VASC) lesions. Our experimental results show that Bayesian deep networks can boost the diagnostic performance of the standard DenseNet-169 model from 81.35% to 83.59% without incurring additional parameters or heavy computation. More importantly, a hybrid physician–machine workflow reaches a classification accuracy of 90 % while only referring 35 % of the cases to physicians. The findings are expected to generalize to other medical diagnosis applications. We believe that the availability of risk-aware machine learning methods will enable a wider adoption of machine learning technology in clinical settings. 

   more » « less
2. Applying virtual histology to reflectance confocal microscopy to distinguish malignant from benign cutaneous squamous neoplasms

   Kokikian, N; Wahhab, R; Li, J; Martin, S; Beynet, D; Ozcan, A; Scumpia, P O (May 2024, Society for Investigative Dermatology (SID) Annual Meeting)

   Reflectance confocal microscopy (RCM) is a noninvasive optical imaging technique that uses a laser to capture cellular-level resolution images based on differing refractive indices of tissue elements. RCM image interpretation is challenging and requires training to interpret and correlate the grayscale output images that lack nuclear features with tissue pathology. Here, we utilize a deep learning-based framework that uses a convolutional neural network to transform grayscale images into virtually-stained hematoxylin and eosin (H&E)-like images enabling the visualization of various skin layers. To train the deep-learning framework, a series of a minimum of 7 time-lapsed, successive “stacks” of RCM images of excised tissue, spaced 1.52μm apart to a depth of 60.96μm were obtained using the Vivascope 1500. The tissue samples were stained with a 50% acetic acid solution to enhance cell nuclei. These images served as the “ground truth” to train a deep convolutional neural network with a conditional generative adversarial network (GAN)-based machine learning algorithm to digitally convert the images into GAN-based H&E-stained digital images. The machine learning algorithm was initially trained and subsequently retrained with new samples, specifically focusing on squamous neoplasms. The trained algorithm was applied to skin lesions that had a clinical differential diagnosis of squamous neoplasms including squamous cell carcinoma, actinic keratosis, seborrheic keratosis, and basal cell carcinoma. Through continuous training and refinement, the algorithm was able to produce high-resolution, histological quality images of different squamous neoplasms. This algorithm may be used in the future to facilitate earlier diagnosis of cutaneous neoplasms and enable greater uptake of noninvasive imaging technology within the medical community. 

   more » « less
3. Improving the Quantitative Analysis of Breast Microcalcifications: A Multiscale Approach

   https://doi.org/10.1007/s10278-022-00751-3  

   Marasinou, Chrysostomos; Li, Bo; Paige, Jeremy; Omigbodun, Akinyinka; Nakhaei, Noor; Hoyt, Anne; Hsu, William (February 2023, Journal of Digital Imaging)

   Abstract Accurate characterization of microcalcifications (MCs) in 2D digital mammography is a necessary step toward reducing the diagnostic uncertainty associated with the callback of indeterminate MCs. Quantitative analysis of MCs can better identify MCs with a higher likelihood of ductal carcinoma in situ or invasive cancer. However, automated identification and segmentation of MCs remain challenging with high false positive rates. We present a two-stage multiscale approach to MC segmentation in 2D full-field digital mammograms (FFDMs) and diagnostic magnification views. Candidate objects are first delineated using blob detection and Hessian analysis. A regression convolutional network, trained to output a function with a higher response near MCs, chooses the objects which constitute actual MCs. The method was trained and validated on 435 screening and diagnostic FFDMs from two separate datasets. We then used our approach to segment MCs on magnification views of 248 cases with amorphous MCs. We modeled the extracted features using gradient tree boosting to classify each case as benign or malignant. Compared to state-of-the-art comparison methods, our approach achieved superior mean intersection over the union (0.670 ± 0.121 per image versus 0.524 ± 0.034 per image), intersection over the union per MC object (0.607 ± 0.250 versus 0.363 ± 0.278) and true positive rate of 0.744 versus 0.581 at 0.4 false positive detections per square centimeter. Features generated using our approach outperformed the comparison method (0.763 versus 0.710 AUC) in distinguishing amorphous calcifications as benign or malignant. 

   more » « less
4. Label-free, in vivo virtual histology of skin using reflectance confocal microscopy and deep learning

   https://doi.org/10.1117/12.2648150  

   Li, Jingxi; Garfinkel, Jason; Zhang, Xiaoran; Wu, Di; Zhang, Yijie; de Haan, Kevin; Wang, Hongda; Liu, Tairan; Bai, Bijie; Rivenson, Yair; et al (March 2023, SPIE Photonics West)

   Shaked, Natan T.; Hayden, Oliver (Ed.)

   We report label-free, in vivo virtual histology of skin using reflectance confocal microscopy (RCM). We trained a deep neural network to transform in vivo RCM images of unstained skin into virtually stained H&E-like microscopic images with nuclear contrast. This framework successfully generalized to diverse skin conditions, e.g., normal skin, basal cell carcinoma, and melanocytic nevi, as well as distinct skin layers, including the epidermis, dermal-epidermal junction, and superficial dermis layers. This label-free in vivo skin virtual histology framework can be transformative for faster and more accurate diagnosis of malignant skin neoplasms, with the potential to significantly reduce unnecessary skin biopsies. 

   more » « less
5. Deep learning-enabled, non-invasive virtual histology of skin using reflectance confocal microscopy

   https://doi.org/10.1117/12.2632602  

   Li, Jingxi; Garfinkel, Jason; Zhang, Xiaoran; Wu, Di; Zhang, Yijie; de Haan, Kevin; Wang, Hongda; Liu, Tairan; Bai, Bijie; Rivenson, Yair; et al (October 2022, SPIE Optics and Photonics Conference)

   Volpe, Giovanni; Pereira, Joana B.; Brunner, Daniel; Ozcan, Aydogan (Ed.)

   Reflectance confocal microscopy (RCM) can provide in vivo images of the skin with cellular-level resolution; however, RCM images are grayscale, lack nuclear features and have a low correlation with histology. We present a deep learning-based virtual staining method to perform non-invasive virtual histology of the skin based on in vivo, label-free RCM images. This virtual histology framework revealed successful inference for various skin conditions, such as basal cell carcinoma, also covering distinct skin layers, including epidermis and dermal-epidermal junction. This method can pave the way for faster and more accurate diagnosis of malignant skin neoplasms while reducing unnecessary biopsies. 

   more » « less