It can take years for patients with rare diseases to get diagnoses and understand more about their conditions. But what if someone developed a new and efficient technology that could hasten the process? This is what Metin Akay, John S. Dunn Endowed Chair Professor of biomedical engineering at the University of Houston (UH), is working to do. In a news release, UH shared Akay’s artificial intelligence (AI) platform with deep neural network architecture. By feeding the network photographs, it can learn to differentiate between systemic sclerosis (SSc) and healthy skin.
For patients with SSc, early diagnosis is key to better treatment and a higher quality of life (QOL). However, it can be difficult to determine disease stage and progression. This deep learning network could provide enhanced diagnostic procedures and fill an unmet need in this patient community. See the research published in the IEEE Open Journal of Engineering in Medicine and Biology.
Akay’s proposal is for a network using a standard laptop computer with a 2.5 Ghz Intel Core i7. He trained the program using a mobile vision application called MobileNetV2. Through MobileNetV2, the program learned approximately 1.4 million images relating to healthy skin and SSc. Through this, the program is able to determine what healthy skin looks like, and can identify any potential abnormalities.
The program is a modified Convolutional Neural Network (CNN). According to data scientist Sumit Saha:
A Convolutional Neural Network (ConvNet/CNN) is a Deep Learning algorithm which can take in an input image, assign importance (learnable weights and biases) to various aspects/objects in the image and be able to differentiate one from the other. The pre-processing required in a ConvNet is much lower as compared to other classification algorithms.
Within the modified structure, training the program only took a little under 5 hours. More so, the program ranged from 95.2-100% accurate on different data sets. This suggests that, with some fine-tuning, this type of CNN program could be incredibly helpful in addressing the difficulties of the current SSc diagnostic process.
Systemic Scleroderma (SSc)
Also known as systemic sclerosis, systemic scleroderma (SSc) is an autoimmune disorder, meaning it is caused by the immune system mistakenly attacking healthy tissue. In this case, SS causes the thickening or hardening of fibrosis (scar tissue) in the skin and organs. This thickening is caused by excess collagen, which normally plays a role in connective tissue strength. Altogether, an estimated 40,000-165,000 Americans have SSc. Females are 4x more likely to develop SSc than males.
Symptoms of SSc include:
- Hair loss
- Joint and muscle pain
- Organ failure
- Skin fibrosis
- Raynaud’s phenomenon (when the fingers and toes turn blue/white in the cold)
- Difficulty swallowing
- Shortness of breath and/or difficulty breathing
- High blood pressure
- Open sores on the hands and fingers
- Painful calcium deposits under the skin
- Kidney issues
- Swelling of the extremities