Comparative Study of Dermoscopic Hair Removal Methods | Semantic Scholar (2024)

This work presents a model based on convolutional neural networks for hair removal in dermoscopic images based on deep learning, and compares its results with six state-of-the-art algorithms based on traditional computer vision techniques by means of similarity measures.

This work presents a new approach for the task of hair removal on dermoscopic images based on deep learning techniques that relies on an encoder-decoder architecture, with convolutional neural networks, for the detection and posterior restoration of hair’s pixels from the images.

An efficient method to enhance dermoscopic images by removing hair and other artifacts using black-hat morphological processing and total variation inpainting technique is proposed and results depict the improvement in classification accuracy of three skin cancer classes, when hair and artifacts are eliminated by proposed method.

The results of the evaluation are promising as the detection of the hair regions is accurate, and the performance results are satisfactory in comparison to other existing hair removal methods.

This work presents a new approach for hair removal on dermoscopic images based on deep learning techniques, as well as study in depth the behavior of the tasks of skin lesion segmentation, hair mask segmentations, and inpainting of those regions, in a multitasking framework to discover how tasks influence each other.

The proposed algorithm uses transfer learning approach which improves classification accuracy with less loss, pre-trained model used are VGG16, VGG19, ResNet50, and these pretrained models have been applied on 3200 dermoscopy skin images taken from ISIC.

This work adopted the multi-layer residual convolutional neural network (MLRNet) for skin cancer segmentation and outperformed in both subjective and objective performances as compared to conventional deep learning approaches.

Simplifying an already light model, such as MobileNetV2, is pursued, combining it with an attention mechanism to enhance the network's capability to learn and compensate for the lack of information that simplifying the original architecture might cause.

Automating the computer‐aided system of skin lesion detection and classification (SLDC) will assist the medical practitioners to ensure more efficacious treatment of skinLesion disease.

...

...

The preliminary results indicated the proposed method was able to remove more fine hairs and hairs in the shade, and lower false hair detection rate by 58% as compared to the DullRazor's approach.

This work provides an effective hair removal algorithm for dermoscopic imagery employing soft color morphology operators able to cope with color images and compares it with other state-of-the-art algorithms.

This work proposes a novel method for simultaneously enhancing both light and dark hairs with variable widths, from dermoscopic images, without the prior knowledge of the hair color, and validate and compare it to other methods.

A new scheme that automatically detects and removes hairs from dermoscopy images is presented that will help in the precise segmentation and analysis of the skin lesions.

A realistic hair simulator based on context-aware image synthesis using image-to-image translation techniques via conditional adversarial generative networks for generation of different hair occlusions in skin images, along with the ground-truth mask for hair location is proposed.

Dermatoscopy appeared not to improve the accuracy of diagnosis enough to alter the clinical management of most pigmented skin lesions and lack of studies in primary care make generalisation of results difficult.

Two new full-reference metrics for image quality assessment based on the Peak-Signal-to-Noise Ratio and Universal Quality Index modified to take into account the Human Visual System (HVS) properties are presented.

...

...