Skin Cancer Detection in Diverse Skin Tones by Machine Learning Combining Audio and Visual Convolutional Neural Networks.

IF 2.5 3区医学 Q3 ONCOLOGY

Oncology Pub Date : 2024-09-23 DOI:10.1159/000541573

Bruce N Walker, Travis Wayne Blalock, Rebecca Leibowitz, Yoram Oron, Daphne Dascalu, Eli Omid David, Avi Dascalu

{"title":"Skin Cancer Detection in Diverse Skin Tones by Machine Learning Combining Audio and Visual Convolutional Neural Networks.","authors":"Bruce N Walker, Travis Wayne Blalock, Rebecca Leibowitz, Yoram Oron, Daphne Dascalu, Eli Omid David, Avi Dascalu","doi":"10.1159/000541573","DOIUrl":null,"url":null,"abstract":"Introduction: Skin cancer (SC) is common in fair skin (FS) at a 1:5 lifetime incidence for nonmelanoma skin cancer. In order to assist clinicians' decisions, a risk intervention technology was developed, which combines a dual-mode machine learning of visual and sonified (pixel to sound) data. The addition of an audio technology enhances malignant features of lesions, increases sensitivity and was previously validated under a prospective clinical setting in FS. In dark skin (DS), although rare by a 10-30 factor, skin cancer is diagnosed at more advanced stages resulting in a delayed diagnosis and affecting life quality and expectancy. It is known as well that SC diagnostic accuracy by machine learning in DS is decreased as compared to FS. The present study tests the use of sonification aided by artificial intelligence algorithms to compare diagnostics of different skin tones.Methodology: Biopsy-validated smartphone images were diagnosed in a retrospective study by a dual audio-visual convoluted neural network. A total of 60 Fitzpatrick I-III were compared to 72 Fitzpatrick IV-VI. A dichotomous diagnostic output, either malignant or benign, was assessed for sensitivity, specificity and area under the curves (AUCs) for the receiver operating characteristic (ROC) curve.Results: ROC curve analytics indicated an AUC of 0.858 (95% CI: 0.795-0.921) and 0.856 (95% CI: 0.759-0.953) for fair and DS (p = NS). Sensitivity of Fitzpatrick I-III skin and Fitzpatrick IV-VI were 84.4% (71.8-96.9) and 79.6% (63.4-93.8), respectively (p = NS). Specificity of Fitzpatrick I-III skin and Fitzpatrick IV-VI were 84.2% (72.6-95.8) and 85.3% (73.4-97.2), respectively (p = NS). The positive predictive and negative predictive values as well as accuracy (0.817 vs. 0.847) were all within the same range (p = NS).Conclusions: The results demonstrate that the dual-modality classifier identifies skin cancer of FS and DS similarly well. Sonification of malignant signs of a skin lesion demonstrates promising results, even with smartphone images, which should be considered as a tool to achieve more effective and accessible healthcare.","PeriodicalId":19497,"journal":{"name":"Oncology","volume":" ","pages":"1-8"},"PeriodicalIF":2.5000,"publicationDate":"2024-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Oncology","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1159/000541573","RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"ONCOLOGY","Score":null,"Total":0}

引用次数: 0

Abstract

Introduction: Skin cancer (SC) is common in fair skin (FS) at a 1:5 lifetime incidence for nonmelanoma skin cancer. In order to assist clinicians' decisions, a risk intervention technology was developed, which combines a dual-mode machine learning of visual and sonified (pixel to sound) data. The addition of an audio technology enhances malignant features of lesions, increases sensitivity and was previously validated under a prospective clinical setting in FS. In dark skin (DS), although rare by a 10-30 factor, skin cancer is diagnosed at more advanced stages resulting in a delayed diagnosis and affecting life quality and expectancy. It is known as well that SC diagnostic accuracy by machine learning in DS is decreased as compared to FS. The present study tests the use of sonification aided by artificial intelligence algorithms to compare diagnostics of different skin tones.

Methodology: Biopsy-validated smartphone images were diagnosed in a retrospective study by a dual audio-visual convoluted neural network. A total of 60 Fitzpatrick I-III were compared to 72 Fitzpatrick IV-VI. A dichotomous diagnostic output, either malignant or benign, was assessed for sensitivity, specificity and area under the curves (AUCs) for the receiver operating characteristic (ROC) curve.

Results: ROC curve analytics indicated an AUC of 0.858 (95% CI: 0.795-0.921) and 0.856 (95% CI: 0.759-0.953) for fair and DS (p = NS). Sensitivity of Fitzpatrick I-III skin and Fitzpatrick IV-VI were 84.4% (71.8-96.9) and 79.6% (63.4-93.8), respectively (p = NS). Specificity of Fitzpatrick I-III skin and Fitzpatrick IV-VI were 84.2% (72.6-95.8) and 85.3% (73.4-97.2), respectively (p = NS). The positive predictive and negative predictive values as well as accuracy (0.817 vs. 0.847) were all within the same range (p = NS).

Conclusions: The results demonstrate that the dual-modality classifier identifies skin cancer of FS and DS similarly well. Sonification of malignant signs of a skin lesion demonstrates promising results, even with smartphone images, which should be considered as a tool to achieve more effective and accessible healthcare.

查看原文本刊更多论文

通过结合音频和视觉卷积神经网络的机器学习，检测不同肤色的皮肤癌。

导言：皮肤癌（SC）常见于皮肤白皙者（FS），非黑色素瘤皮肤癌的终生发病率为 1:5。为了帮助临床医生做出决策，我们开发了一种风险干预技术，该技术结合了视觉和声音（像素到声音）数据的双模式机器学习。音频技术的加入增强了病变的恶性特征，提高了灵敏度，并已在 FS 的前瞻性临床环境中得到验证。深色皮肤（DS）的皮肤癌虽然罕见，但其发病率为 10-30 倍，而且皮肤癌的诊断处于晚期，导致诊断延迟，影响生活质量和预期寿命。众所周知，与 FS 相比，机器学习对 DS 的 SC 诊断准确率有所下降。本研究测试了在人工智能算法辅助下使用超声波来比较不同肤色的诊断结果：方法：在一项回顾性研究中，通过双视听卷积神经网络对经活检验证的智能手机图像进行诊断。共有 60 张 Fitzpatrick I-III 与 72 张 Fitzpatrick IV-VI 进行了比较。对二分诊断结果（恶性或良性）的敏感性、特异性和接收者操作特征曲线（ROC）的曲线下面积（AUC）进行了评估：ROC曲线分析表明，公平和DS的AUC分别为0.858（95% CI 0.795-0.921）和0.856（95% CI 0.759-0.953）（P=NS）。Fitzpatrick I-III 皮肤和 Fitzpatrick IV-VI 皮肤的敏感性分别为 84.4%（71.8-96.9）和 79.6%（63.4-93.8）（p=NS）。Fitzpatrick I-III 皮肤和 Fitzpatrick IV-VI 的特异性分别为 84.2%（72.6 至 95.8）和 85.3%（73.4 至 97.2）（P=NS）。阳性预测值和阴性预测值以及准确度（0.817 对 0.847）均在同一范围内（P=NS）：结果表明，双模态分类器对 FS 和 DS 皮肤癌的识别效果类似。即使使用智能手机图像，皮肤病变恶性征兆的声纳化也能显示出良好的效果，应将其视为实现更有效、更方便的医疗保健的一种工具。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Oncology 医学-肿瘤学

CiteScore

6.00

自引率

2.90%

发文量

审稿时长

6-12 weeks

期刊介绍： Although laboratory and clinical cancer research need to be closely linked, observations at the basic level often remain removed from medical applications. This journal works to accelerate the translation of experimental results into the clinic, and back again into the laboratory for further investigation. The fundamental purpose of this effort is to advance clinically-relevant knowledge of cancer, and improve the outcome of prevention, diagnosis and treatment of malignant disease. The journal publishes significant clinical studies from cancer programs around the world, along with important translational laboratory findings, mini-reviews (invited and submitted) and in-depth discussions of evolving and controversial topics in the oncology arena. A unique feature of the journal is a new section which focuses on rapid peer-review and subsequent publication of short reports of phase 1 and phase 2 clinical cancer trials, with a goal of insuring that high-quality clinical cancer research quickly enters the public domain, regardless of the trial’s ultimate conclusions regarding efficacy or toxicity.