Rahmetullah Varol

Abdurrahim Yilmaz, Furkan Yuceyalcin, Ece Gokyayla et al.

A major barrier to developing vision large language models (LLMs) in dermatology is the lack of large image--text pairs dataset. We introduce DermaSynth, a dataset comprising of 92,020 synthetic image--text pairs curated from 45,205 images (13,568 clinical and 35,561 dermatoscopic) for dermatology-related clinical tasks. Leveraging state-of-the-art LLMs, using Gemini 2.0, we used clinically related prompts and self-instruct method to generate diverse and rich synthetic texts. Metadata of the datasets were incorporated into the input prompts by targeting to reduce potential hallucinations. The resulting dataset builds upon open access dermatological image repositories (DERM12345, BCN20000, PAD-UFES-20, SCIN, and HIBA) that have permissive CC-BY-4.0 licenses. We also fine-tuned a preliminary Llama-3.2-11B-Vision-Instruct model, DermatoLlama 1.0, on 5,000 samples. We anticipate this dataset to support and accelerate AI research in dermatology. Data and code underlying this work are accessible at https://github.com/abdurrahimyilmaz/DermaSynth.

Abdurrahim Yilmaz, Serra Atilla Aydin, Deniz Temur et al.

Mohs micrographic surgery (MMS) is the gold standard technique for removing high risk nonmelanoma skin cancer however, intraoperative histopathological examination demands significant time, effort, and professionality. The objective of this study is to develop a deep learning model to detect basal cell carcinoma (BCC) and artifacts on Mohs slides. A total of 731 Mohs slides from 51 patients with BCCs were used in this study, with 91 containing tumor and 640 without tumor which was defined as non-tumor. The dataset was employed to train U-Net based models that segment tumor and non-tumor regions on the slides. The segmented patches were classified as tumor, or non-tumor to produce predictions for whole slide images (WSIs). For the segmentation phase, the deep learning model success was measured using a Dice score with 0.70 and 0.67 value, area under the curve (AUC) score with 0.98 and 0.96 for tumor and non-tumor, respectively. For the tumor classification, an AUC of 0.98 for patch-based detection, and AUC of 0.91 for slide-based detection was obtained on the test dataset. We present an AI system that can detect tumors and non-tumors in Mohs slides with high success. Deep learning can aid Mohs surgeons and dermatopathologists in making more accurate decisions.

Abdurrahim Yilmaz, Fatih Goktay, Rahmetullah Varol et al.

The diagnosis of superficial fungal infections in dermatology is still mostly based on manual direct microscopic examination with Potassium Hydroxide (KOH) solution. However, this method can be time consuming and its diagnostic accuracy rates vary widely depending on the clinician's experience. With the increase of neural network applications in the field of clinical microscopy, it is now possible to automate such manual processes increasing both efficiency and accuracy. This study presents a deep neural network structure that enables the rapid solutions for these problems and can perform automatic fungi detection in grayscale images without dyes. 160 microscopic field photographs containing the fungal element, obtained from patients with onychomycosis, and 297 microscopic field photographs containing dissolved keratin obtained from normal nails were collected. Smaller patches containing 4234 fungi and 4981 keratin were extracted from these images. In order to detect fungus and keratin, VGG16 and InceptionV3 models were developed. The VGG16 model had 95.98% accuracy, and the area under the curve (AUC) value of 0.9930, while the InceptionV3 model had 95.90% accuracy and the AUC value of 0.9917. However, average accuracy and AUC value of clinicians is 72.8% and 0.87, respectively. This deep learning model allows the development of an automated system that can detect fungi within microscopic images.

Rahmetullah Varol, Sevde Omeroglu, Zeynep Karavelioglu et al.

Accurate assessment of stiffness distribution is essential due to the critical role of single cell mechanobiology in the regulation of many vital cellular processes such as proliferation, adhesion, migration, and motility. Cell stiffness is one of the fundamental mechanical properties of the cell and is greatly affected by the intracellular tensional forces, cytoskeletal prestress, and cytoskeleton structure. Herein, we propose a novel holographic single-cell stiffness measurement technique that can obtain the stiffness distribution over a cell membrane at high resolution and in real-time. The proposed imaging method coupled with acoustic signals allows us to assess the cell stiffness distribution with a low error margin and label-free manner. We demonstrate the proposed technique on HCT116 (Human Colorectal Carcinoma) cells and CTC-mimicked HCT116 cells by induction with transforming growth factor-beta (TGF-\b{eta}). Validation studies of the proposed approach were carried out on certified polystyrene microbeads with known stiffness levels. Its performance was evaluated in comparison with the AFM results obtained for the relevant cells. When the experimental results were examined, the proposed methodology shows utmost performance over average cell stiffness values for HCT116, and CTC-mimicked HCT116 cells were found as 1.08 kPa, and 0.88 kPa, respectively. The results confirm that CTC-mimicked HCT116 cells lose their adhesion ability to enter the vascular circulation and metastasize. They also exhibit a softer stiffness profile compared to adherent forms of the cancer cells. Hence, the proposed technique is a significant, reliable, and faster alternative for in-vitro cell stiffness characterization tools. It can be utilized for various applications where single-cell analysis is required, such as disease modeling, drug testing, diagnostics, and many more.