Early diagnosis of skin cancer by ultrasound frequency analysis

Abstract

The diagnosis of cancer by modern computer tools, at the very first stages of the incident, is a very important issue that has involved many researchers. In the meantime, skin cancer is a great deal of research because many people are involved with it. The purpose of this paper is to introduce an innovative method based on tissue frequency analyzes to obtain the accurate and real‐time evaluation of skin cancers. According to the Biological resonance theory, body cells have natural and unique frequencies based on their biological fluctuations, which, if the structure, profile and cellular status change, its frequency also varies. This concept and theory is considered as the basis for analyzing skin tissue health in the proposed method. Reflected ultrasound waves from tissue have been processed and studied based on frequency analysis as a new method for early detection and diagnosis of accurate location and type of skin diseases. The developed algorithm was approved through 400 patients from CRED; its ability to evaluate benign and malignant skin lesions was shown (AUC = 0.959), with comparable clinical precision; as for the selected threshold, sensitivity, and specificity were 93.8% and 97.3%, respectively. Therefore, this method can detect skin malignancy with an accurate, noninvasive and real‐time procedure.

Computer-Aided Diagnosis Methods for High-Frequency Ultrasound Data Analysis: A Review

“…For 91 multimodal samples, the accuracy of the melanocytic lesion classification (malignant melanoma and melanocytic nevi) was equal to 0.989 and the AUC was close to one. For the analysis of single modality (400 images, 50 MHz), noteworthy results (AUC equal to 0.917) were reported by [ 45 ] in the four-group classification: healthy skin, benign nevi, BCC, and melanoma. For the two-group analysis (310 benign and malignant lesions), the best solution was provided in [ 21 ], with an AUC equal to 0.953.…”

“…For the analysis of single modality (400 images, 50 MHz), noteworthy results (AUC equal to 0.917) were reported by [ 45 ] in the four-group classification: healthy skin, benign nevi, BCC, and melanoma. For the two-group analysis (310 benign and malignant lesions), the best solution was provided in [ 21 ], with an AUC equal to 0.953.…”

“…The second most widely explored area was skin tumor segmentation [ 20 , 26 , 63 , 75 , 76 , 78 ]. In support of skin tumor diagnosis, there also appeared works targeting HFUS image classification [ 21 , 45 , 46 , 81 , 82 , 83 ], with a particular emphasis on melanocytic lesion classification. However, other classification problems were also found in the literature [ 74 , 86 ].…”

“…Three years later, Kia et al [ 45 ] developed a skin image classification algorithm based on a discrete cosine transform (DCT), followed by a singular value decomposition (SVD) and neural network with two hidden layers (10 neurons each) and a CE loss function. The extended (compared to [ 81 ]) database of 400 HFUS samples included melanomas, basal cell carcinomas, squamous cell carcinomas, actinic keratosis, atypical nevi, benign melanocytic nevi, blue nevi, and seborrheic keratosis.…”

Automated Skin Lesion Classification on Ultrasound Images

“…In addition, healthy skin without lesions was included in the testing set, making a comparison with other classification articles difficult, if they do not consider the differentiation of lesion-free skin necessary. A more recent work from the group is based on tissue frequency analysis [ 51 ]. It uses a 384-element-long feature vector from frequency space to train the above-mentioned neural network.…”