Alshardan Amal, Saeed Muhammad Kashif, Alotaibi Shoayee Dlaim, Alashjaee Abdullah M, Salih Nahla, Marzouk Radwa
Department of Computer Science, College of Computer and Information Sciences, Princess Nourah bint Abdulrahman University (PNU), P.O. Box 84428, 11671 Riyadh, Saudi Arabia.
Department of Computer Science, Applied College at Mahayil, King Khalid University, Abha, Saudi Arabia.
Health Inf Sci Syst. 2024 May 15;12(1):35. doi: 10.1007/s13755-024-00294-7. eCollection 2024 Dec.
Gastrointestinal (GI) cancer detection includes the detection of cancerous or potentially cancerous lesions within the GI tract. Earlier diagnosis is critical for increasing the success of treatment and improving patient outcomes. Medical imaging plays a major role in diagnosing and detecting GI cancer. CT scans, endoscopy, MRI, ultrasound, and positron emission tomography (PET) scans can help detect lesions, abnormal masses, and changes in tissue structure. Artificial intelligence (AI) and machine learning (ML) methods are being gradually applied to medical imaging for cancer diagnosis. ML algorithms, including deep learning methodologies like convolutional neural network (CNN), are applied frequently for cancer diagnosis. These models learn features and patterns from labelled datasets to discriminate between normal and abnormal areas in medical images. This article presents a new Harbor Seal Whiskers Optimization Algorithm with Deep Learning based Medical Imaging Analysis for Gastrointestinal Cancer Detection (HSWOA-DLGCD) technique. The goal of the HSWOA-DLGCD algorithm is to explore the GI images for the cancer diagnosis. In order to accomplish this, the HSWOA-DLGCD system applies bilateral filtering (BF) approach for the removal of noise. In addition, the HSWOA-DLGCD technique makes use of HSWOA with Xception model for feature extraction. For cancer recognition, the HSWOA-DLGCD technique applies extreme gradient boosting (XGBoost) model. Finally, the parameters compared with the XGBoost system can be selected by moth flame optimization (MFO) system. The experimental results of the HSWOA-DLGCD technique could be verified on the Kvasir database. The simulation outcome demonstrated a best possible solution of the HSWOA-DLGCD method than other recent methods.
胃肠道（GI）癌症检测包括检测胃肠道内的癌性或潜在癌性病变。早期诊断对于提高治疗成功率和改善患者预后至关重要。医学成像在胃肠道癌症的诊断和检测中起着主要作用。计算机断层扫描（CT）、内窥镜检查、磁共振成像（MRI）、超声和正电子发射断层扫描（PET）有助于检测病变、异常肿块和组织结构变化。人工智能（AI）和机器学习（ML）方法正逐渐应用于医学成像以进行癌症诊断。包括卷积神经网络（CNN）等深度学习方法在内的ML算法经常用于癌症诊断。这些模型从标记数据集中学习特征和模式，以区分医学图像中的正常区域和异常区域。本文提出了一种基于深度学习的医学成像分析的新型斑海豹须优化算法用于胃肠道癌症检测（HSWOA-DLGCD）技术。HSWOA-DLGCD算法的目标是探索胃肠道图像以进行癌症诊断。为了实现这一目标，HSWOA-DLGCD系统应用双边滤波（BF）方法去除噪声。此外，HSWOA-DLGCD技术利用HSWOA与Xception模型进行特征提取。对于癌症识别，HSWOA-DLGCD技术应用极端梯度提升（XGBoost）模型。最后，可以通过蛾火焰优化（MFO）系统选择与XGBoost系统相比的参数。HSWOA-DLGCD技术的实验结果可以在Kvasir数据库上得到验证。模拟结果表明，HSWOA-DLGCD方法比其他最近的方法具有更好的解决方案。
