Automated Deep Learning Based Melanoma Detection and Classification Using Biomedical Dermoscopic Images

Abstract

Melanoma remains a serious illness which is a common form of skin cancer. Since the earlier detection of melanoma reduces the mortality rate, it is essential to design reliable and automated disease diagnosis model using dermoscopic images. The recent advances in deep learning (DL) models find useful to examine the medical image and make proper decisions. In this study, an automated deep learning based melanoma detection and classification (ADL-MDC) model is presented. The goal of the ADL-MDC technique is to examine the dermoscopic images to determine the existence of melanoma. The ADL-MDC technique performs contrast enhancement and data augmentation at the initial stage. Besides, the k-means clustering technique is applied for the image segmentation process. In addition, Adagrad optimizer based Capsule Network (CapsNet) model is derived for effective feature extraction process. Lastly, crow search optimization (CSO) algorithm with sparse autoencoder (SAE) model is utilized for the melanoma classification process. The exploitation of the Adagrad and CSO algorithm helps to properly accomplish improved performance. A wide range of simulation analyses is carried out on benchmark datasets and the results are inspected under several aspects. The simulation results reported the enhanced performance of the ADL-MDC technique over the recent approaches.

Quantum computing meets skin cancer diagnosis

“…This means most models should achieve these values to perform on par with dermatologists and aim to surpass them. Nevertheless, computer vision classification relies on the implemented approach for the task in hand, as it can be seen in Albraikan's work, 14 where high performance in multiclass classification is achieved through a novel automated deep learning technique comprised of several stages for different split ratios of the ISIC dataset. However, one key aspect in model performance is the number of samples in the training dataset and the difficulty for feature abstraction, as shown by, 15 where multiclass classification for the HAM1000 dataset is not on par with, 14 which can be mainly attributed to the dataset characteristics.…”

“…Nevertheless, computer vision classification relies on the implemented approach for the task in hand, as it can be seen in Albraikan's work, 14 where high performance in multiclass classification is achieved through a novel automated deep learning technique comprised of several stages for different split ratios of the ISIC dataset. However, one key aspect in model performance is the number of samples in the training dataset and the difficulty for feature abstraction, as shown by, 15 where multiclass classification for the HAM1000 dataset is not on par with, 14 which can be mainly attributed to the dataset characteristics. For binary classification, 11 and 12 show the capability of deep learning for the task of melanoma classification, in which different preprocessing techniques aid the implemented models to attain higher performance than physicians for the ISIC dataset as seen in Brinker's work.…”

“…In 2023, a novel Automated Deep Learning Melanoma Detection and Classification (ADL-MDC) technique, which includes preprocessing, K-means segmentation, feature extraction, and parameter optimization, obtained better results over recent deep learning methods. 14 Further recent works focus on testing different approaches to the task via modifying the training data split, 15 as well as implementing ensemble CNN methods, 16 and innovative frameworks capable of improved feature extraction. 17 In contrast, quantum computing and Quantum Machine Learning (QML) are making advancements in the possible problems to solve through these techniques, ranging from time series to medical image classification, where quantum hybrid algorithms attempt to achieve the performance of classical models considering different conditions, and in some cases surpass them.…”

Efficient Brain Tumor Detection with Lightweight End-to-End Deep Learning Model

“…Traditional computer vision techniques for brain tumor detection often involve manual feature extraction and classification using techniques such as support vector machines (SVM), decision trees, and Random Forest [23,24]. In recent years, deep learning techniques such as convolutional neural networks (CNNs) have become increasingly popular for brain tumor detection due to their ability to automatically learn features from medical imaging data [25][26][27]. These models have shown promising results in detecting and segmenting brain tumors, as well as in classifying different types of tumors.…”

“…The first step involves importing the TensorFlow library as tf, which is a popular open-source software library for building machine learning models [26]. Next, a sequential model architecture is defined using the tf.keras.Sequential method.…”