Minimal preprocessing improves skin lesion analysis with fast small model
Less Is More: A Leakage-Controlled Study of Dermoscopic Preprocessing for Joint Skin Lesion Classification and Segmentation with YOLO26
Machine LearningComputer Vision and Pattern Recognition
Summary
Images of skin lesions often have artifacts that can confuse automatic analysis. This study tested whether complex image cleaning steps actually help a small, fast AI model that identifies and outlines different types of skin lesions. The researchers found that keeping images mostly natural and using simple, standard image tweaks during training worked better than using more complex pre-processing methods. Their setup was carefully designed to avoid confusing the model with repeated images of the same lesion, making the results more trustworthy.
What this means in practice
- •For medical imaging developers: Build efficient skin lesion analysis systems using minimal image preprocessing for better accuracy and speed.
- •For mobile health app developers: Implement lightweight, real-time dermoscopic lesion detection and segmentation using fast YOLO26n models with simple image inputs.$Commercial implications: Enables development of fast, on-device skin lesion screening apps that run efficiently on mobile devices.
Tested on one dataset.
Authors
Truong Viet Vu, Nguyen Chi Hai, Nguyen Phuc Nguyen, Ngo Hoang Tu, Vo Nguyen Quoc Bao, Nguyen Thai Anh
Abstract
Handcrafted preprocessing is widely employed in automated dermoscopic analysis to suppress imaging artifacts and enhance lesion visibility. Nevertheless, its actual contribution to modern real-time models remains unclear, particularly when evaluation protocols do not adequately control correlations among images of the same lesion. This study presents a leakage-controlled, lesion-disjoint evaluation of dermoscopic preprocessing and augmentation for joint multi-class lesion classification and instance segmentation using a fixed nano-scale YOLO26 segmentation model (YOLO26n-seg). From HAM10000 (10,015 images), quality control yields 10,013 valid image-mask pairs from 7,468 unique lesions, partitioned into mutually exclusive sets by lesion identity. With the architecture, resolution, training budget, and evaluation protocol held fixed, we compare minimally processed images plus online augmentation against offline class balancing, DullRazor-CLAHE preprocessing, and raw-processed hybrid views, over three random seeds. On the lesion-disjoint test set, the raw baseline achieves a mask mAP$_{50:95}$ of $0.5636 \pm 0.0234$, a Dice score of $0.9356 \pm 0.0024$, and a macro-F1 score of $0.6917 \pm 0.0202$. Offline augmentation does not improve the mean performance, while the combined and hybrid strategies reduce both class-aware segmentation and classification accuracy. At only 2.69 million parameters, the model runs at approximately 50 frames per second. Under a leakage-controlled, lesion-disjoint protocol with all non-input factors held fixed, minimally processed dermoscopic images combined with standard online augmentation deliver a better accuracy-efficiency trade-off than increasingly complex deterministic preprocessing, which yields no consistent joint benefit across three seeds on HAM10000.