A survey on deep learning in medical image analysis

Deep convolutional neural networks have performed remarkably well on many Computer Vision tasks. However, these networks are heavily reliant on big data to avoid overfitting. Overfitting refers to the phenomenon when a network learns a function with very high variance such as to perfectly model the training data. Unfortunately, many application domains do not have access to big data, such as medical image analysis. This survey focuses on Data Augmentation, a data-space solution to the problem of limited data. Data Augmentation encompasses a suite of techniques that enhance the size and quality of training datasets such that better Deep Learning models can be built using them. The image augmentation algorithms discussed in this survey include geometric transformations, color space augmentations, kernel filters, mixing images, random erasing, feature space augmentation, adversarial training, generative adversarial networks, neural style transfer, and meta-learning. The application of augmentation methods based on GANs are heavily covered in this survey. In addition to augmentation techniques, this paper will briefly discuss other characteristics of Data Augmentation such as test-time augmentation, resolution impact, final dataset size, and curriculum learning. This survey will present existing methods for Data Augmentation, promising developments, and meta-level decisions for implementing Data Augmentation. Readers will understand how Data Augmentation can improve the performance of their models and expand limited datasets to take advantage of the capabilities of big data.

/pdf/a-survey-on-image-data-augmentation-for-deep-learning-4ru1bj2hbb.pdf

A survey on Image Data Augmentation for Deep Learning

The use of artificial intelligence, and the deep-learning subtype in particular, has been enabled by the use of labeled big data, along with markedly enhanced computing power and cloud storage, across all sectors. In medicine, this is beginning to have an impact at three levels: for clinicians, predominantly via rapid, accurate image interpretation; for health systems, by improving workflow and the potential for reducing medical errors; and for patients, by enabling them to process their own data to promote health. The current limitations, including bias, privacy and security, and lack of transparency, along with the future directions of these applications will be discussed in this article. Over time, marked improvements in accuracy, productivity, and workflow will likely be actualized, but whether that will be used to improve the patient-doctor relationship or facilitate its erosion remains to be seen.

High-performance medicine: the convergence of human and artificial intelligence

The paper “Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification” by Joy Buolamwini and Timnit Gebru, that will be presented at the Conference on Fairness, Accountability, and Transparency (FAT*) in February 2018, evaluates three commercial API-based classifiers of gender from facial images, including IBM Watson Visual Recognition. The study finds these services to have recognition capabilities that are not balanced over genders and skin tones [1]. In particular, the authors show that the highest error involves images of dark-skinned women, while the most accurate result is for light-skinned men.

/pdf/gender-shades-intersectional-accuracy-disparities-in-4qgeu0c1i3.pdf

Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification

Deep neural networks (DNNs) are currently widely used for many artificial intelligence (AI) applications including computer vision, speech recognition, and robotics. While DNNs deliver state-of-the-art accuracy on many AI tasks, it comes at the cost of high computational complexity. Accordingly, techniques that enable efficient processing of DNNs to improve energy efficiency and throughput without sacrificing application accuracy or increasing hardware cost are critical to the wide deployment of DNNs in AI systems. This article aims to provide a comprehensive tutorial and survey about the recent advances toward the goal of enabling efficient processing of DNNs. Specifically, it will provide an overview of DNNs, discuss various hardware platforms and architectures that support DNNs, and highlight key trends in reducing the computation cost of DNNs either solely via hardware design changes or via joint hardware design and DNN algorithm changes. It will also summarize various development resources that enable researchers and practitioners to quickly get started in this field, and highlight important benchmarking metrics and design considerations that should be used for evaluating the rapidly growing number of DNN hardware designs, optionally including algorithmic codesigns, being proposed in academia and industry. The reader will take away the following concepts from this article: understand the key design considerations for DNNs; be able to evaluate different DNN hardware implementations with benchmarks and comparison metrics; understand the tradeoffs between various hardware architectures and platforms; be able to evaluate the utility of various DNN design techniques for efficient processing; and understand recent implementation trends and opportunities.

Efficient Processing of Deep Neural Networks: A Tutorial and Survey

Dupilumab treatment of nummular dermatitis: A retrospective cohort study

Telehealth is an increasingly critical component of the health care ecosystem, especially due to the COVID-19 pandemic. Rapid adoption of telehealth has exposed limitations in the existing infrastructure. In this paper, we study and highlight photo quality as a major challenge in the telehealth workflow. We focus on teledermatology, where photo quality is particularly important; the framework proposed here can be generalized to other health domains. For telemedicine, dermatologists request that patients submit images of their lesions for assessment. However, these images are often of insufficient quality to make a clinical diagnosis since patients do not have experience taking clinical photos. A clinician has to manually triage poor quality images and request new images to be submitted, leading to wasted time for both the clinician and the patient. We propose an automated image assessment machine learning pipeline, TrueImage, to detect poor quality dermatology photos and to guide patients in taking better photos. Our experiments indicate that TrueImage can reject 50% of the sub-par quality images, while retaining 80% of good quality images patients send in, despite heterogeneity and limitations in the training data. These promising results suggest that our solution is feasible and can improve the quality of teledermatology care.

TrueImage: A Machine Learning Algorithm to Improve the Quality of Telehealth Photos

Telehealth is an increasingly critical component of the health care ecosystem, especially due to the COVID-19 pandemic. Rapid adoption of telehealth has exposed limitations in the existing infrastructure. In this paper, we study and highlight photo quality as a major challenge in the telehealth workflow. We focus on teledermatology, where photo quality is particularly important; the framework proposed here can be generalized to other health domains. For telemedicine, dermatologists request that patients submit images of their lesions for assessment. However, these images are often of insufficient quality to make a clinical diagnosis since patients do not have experience taking clinical photos. A clinician has to manually triage poor quality images and request new images to be submitted, leading to wasted time for both the clinician and the patient. We propose an automated image assessment machine learning pipeline, TrueImage, to detect poor quality dermatology photos and to guide patients in taking better photos. Our experiments indicate that TrueImage can reject ~50% of the sub-par quality images, while retaining ~80% of good quality images patients send in, despite heterogeneity and limitations in the training data. These promising results suggest that our solution is feasible and can improve the quality of teledermatology care.

TrueImage: A Machine Learning Algorithm to Improve the Quality of Telehealth Photos.

Question: In outpatients who present with new cutaneous warts, what is the efficacy of treatment with cryotherapy, topical salicylic acid, and a wait-and-see approach in leading to clearance of baseline lesions at 13 weeks? Design: Unblinded, randomized-controlled trial. Setting: Multipractice, 30 family practices from the Leiden Primary Care Research Network in the Netherlands. Patients: Patients 4 years and older who presented to the general medicine clinic with one or more new cutaneous warts. Immunocompromised patients, those with genital warts, seborrheic warts, or warts larger than 1cm in diameter were excluded from the study. Intervention: Enrolled participants were randomized to 13-week treatment periods with cryotherapy (one session every two weeks until resolution of the wart), salicylic acid (daily self-application of 40% salicylic acid in petroleum jelly daily under occlusion preceded by gentle paring), or wait-and-see (advised not to undergo treatment other than over-the-counter medication). Patients were stratified according to characteristics, notably location of warts (plantar vs common [those with both common and plantar warts were grouped according to where the majority of their warts were located]) and number of warts ( 6 or 6). Outcome: The primary end point was the proportion of patients whose warts present at baseline were all cured at 13 weeks. Secondary analyses and subgroup analyses were predefined. Results: Of 250 randomized patients who received treatment, 10 were lost to follow-up at 13 weeks, and 48 patients stopped the assigned treatment protocol. The intention-to-treat analysis demonstrated cure rates of 39% (30/76) for the cryotherapy arm, 24% (20/82) for the salicylic acid arm, and 16% (13/82) for the wait-and-seen arm; two-sided 2 test, linear-by-linear association, comparing three treatment groups (P=.001). With subgroup analysis based on location of warts, cure rates of 49% (19/39) for cryotherapy, 15% (6/39) for salicylic acid, and 8% (3/38) for wait-and see were demonstrated for common warts; two-sided 2 test, linearby-linear association, comparing three treatment groups (P .001). For plantar warts, cure rates of 30% (11/37) for cryotherapy, 33% (14/43) for salicylic acid, and 23% (10/44) for wait and see two-sided 2 test, linear-bylinear association, comparing three treatment groups (P=.46) were seen. Secondary analyses including 26-week follow-up was performed (after allowing for switching after 13 weeks) and results were noted to be concordant with results at 13 weeks. Likewise, further analyses were performed: exclusion of participants with both common and plantar warts from the analysis, per-protocol analysis, and analysis based on individual warts as the unit rather than patients. All of these secondary analyses showed the same significant results as the primary analysis. Also of note, in the common wart group, patients were more satisfied after treatment with cryotherapy, 69%, compared with 24% after salicylic acid and 22% after wait-and-see (P .001). Assessment of adverse effects showed significantly greater number of adverse effects with cryotherapy than with salicylic acid, including more pain and blistering. When comparing the burden of treatment, fewer patients with common warts reported considerable treatment burden with cryotherapy, 31%, as compared with salicylic acid, 49%, two-sided 2 test, linearby-linear association, comparing two treatment groups (P=.04). No difference in patient satisfaction or treatment burden was seen in the plantar wart group. Conclusions: Use of cryotherapy was significantly more effective than use of salicylic acid treatment for common warts only. For plantar warts, there was no clinically relevant difference in effect favoring the use of cryotherapy, salicylic acid, or a wait-and-see approach.

Randomized Controlled Trial of Cryotherapy With Liquid Nitrogen vs Topical Salicylic Acid vs Wait-and-See for Cutaneous Warts

Background Both clinicians and patients have increasingly turned to telemedicine to improve care access, even in physical examination–dependent specialties such as dermatology. However, little is known about whether teledermatology supports effective and timely transitions from inpatient to outpatient care, which is a common care coordination gap. Objective Using mixed methods, this study sought to retrospectively evaluate how teledermatology affected clinic capacity, scheduling efficiency, and timeliness of follow-up care for patients transitioning from inpatient to outpatient dermatology care. Methods Patient-level encounter scheduling data were used to compare the number and proportion of patients who were scheduled and received in-clinic or video dermatology follow-ups within 14 and 90 days after discharge across 3 phases: June to September 2019 (before teledermatology), June to September 2020 (early teledermatology), and February to May 2021 (sustained teledermatology). The time from discharge to scheduling and completion of patient follow-up visits for each care modality was also compared. Dermatology clinicians and schedulers were also interviewed between April and May 2021 to assess their perceptions of teledermatology for postdischarge patients. Results More patients completed follow-up within 90 days after discharge during early (n=101) and sustained (n=100) teledermatology use than at baseline (n=74). Thus, the clinic’s capacity to provide follow-up to patients transitioning from inpatient increased from baseline by 36% in the early (101 from 74) and sustained (100 from 74) teledermatology periods. During early teledermatology use, 61.4% (62/101) of the follow-ups were conducted via video. This decreased significantly to 47% (47/100) in the following year, when COVID-19–related restrictions started to lift (P=.04), indicating more targeted but still substantial use. The proportion of patients who were followed up within the recommended 14 days after discharge did not differ significantly between video and in-clinic visits during the early (33/62, 53% vs 15/39, 38%; P=.15) or sustained (26/53, 60% vs 28/47, 49%; P=.29) teledermatology periods. Interviewees agreed that teledermatology would continue to be offered. Most considered postdischarge follow-up patients to be ideal candidates for teledermatology as they had undergone a recent in-person assessment and might have difficulty attending in-clinic visits because of competing health priorities. Some reported patients needing technological support. Ultimately, most agreed that the choice of follow-up care modality should be the patient’s own. Conclusions Teledermatology could be an important tool for maintaining accessible, flexible, and convenient care for recently discharged patients needing follow-up care. Teledermatology increased clinic capacity, even during the pandemic, although the timeliness of care transitions did not improve. Ultimately, the care modality should be determined through communication with patients to incorporate their and their caregivers’ preferences.

/pdf/teledermatology-to-facilitate-patient-care-transitions-from-2qwq2l2f.pdf

Justin M. Ko

Papers

Dupilumab treatment of nummular dermatitis: A retrospective cohort study

TrueImage: A Machine Learning Algorithm to Improve the Quality of Telehealth Photos

TrueImage: A Machine Learning Algorithm to Improve the Quality of Telehealth Photos.

Randomized Controlled Trial of Cryotherapy With Liquid Nitrogen vs Topical Salicylic Acid vs Wait-and-See for Cutaneous Warts

Teledermatology to Facilitate Patient Care Transitions From Inpatient to Outpatient Dermatology: Mixed Methods Evaluation