Quantifying Alopecia Areata Using Computational Image Analysis

Year: 
2016
PI Name: 
Leslie Castelo-Soccio, MD, PhD
Type:
Pilot & Feasibility Grant
Status: 
Complete

Summary

The purpose of this project is to create an image analysis software platform using image processing and objective algorithms for dermatology applications that can be easily used in a clinician’s or investigator’s office in real time. 

Abstract

In alopecia areata, how dermatologists perceive hair density and skin appearance beyond simple area estimates is extremely difficult. This task is critical when standardizing treatment for patients, following patients from visit to visit, and for testing new medications. Currently most physicians measure the area or take a simple photo but do not further quantify the hair number, type (e.g. vellus, exclamation point, terminal), or color. Drs. Hordinsky, Olsen, Price and colleagues have created guidelines using percent area or the SALT (Severity of Alopecia Tool) score and Visual Analog Scores to move forward clinical research. Our goal is to create a database of images from patients over time to refine a quantitative system for assessing alopecia through automated image analysis. This will provide for the first time truly quantitative answers to the important question that patients and their parents invariably ask: is the alopecia improving or worsening and how can you tell? This method could become a standard of practice, initially for research and ultimately for routine clinical care, accessible to a wide range of dermatologists. Our goals are three fold: first, become a second set of eyes for physicians who see change intuitively but cannot quantify it; second, identify new descriptors of alopecia progression/improvement to be used for assessing alopecia areata; and finally, use this information for standardizing clinical trials.

Impact

If successful, this study could lead to improved prognostic assessments, treatment standardization, and improved patient care for ongoing clinical research and evalutaion in clinics.

Publications

  • Bernardis E, Castelo-Soccio L. Quantifying Alopecia Areata via Texture Analysis to Automate the SALT Score Computation. J Investig Dermatol Symp Proc. 2018 Jan;19(1):S34-S40. doi: 10.1016/j.jisp.2017.10.010.
  • Bernardis E, Nukpezah J, Li P, Christensen T, Castelo-Soccio L. Pediatric severity of alopecia tool. Pediatr Dermatol. 2018 Jan;35(1):e68-e69. doi: 10.1111/pde.13327. Epub 2017 Nov 6.