Liquid based cytology pap smear images for multi-class diagnosis of cervical cancer

While a publicly available benchmark dataset provides a base for the development of new algorithms and comparison of results, hospital-based data collected from the real-world clinical setup is also very important in automated AI-based medical research likewise in disease diagnosis or categorization of predicted disease for tissue level staging or any class identification as per standard protocol so that the developed algorithm works with as much accuracy as possible in the regional context. The repository supports research work related to image segmentation and final classification for a complete decision support system. Liquid based cytology is one of the cervical screening tests. The repository consists of total 963 images sub-divided into four sets of images representing the four classes of pre-cancerous and cancerous lesions of cervical cancer as per standards under The Bethesda System. The pap smear images were captured in 40x magnification using Leica ICC50 HD microscope which is collected and prepared using the liquid-based cytology technique from 460 patients. Microscopic investigation of abnormal changes in cell-level enables detection of malignancy or pre-malignant characteristics. This procedure is time-consuming and subject to inter or intra-observer variability which is why computer-assisted diagnosis can improve the overall disease diagnosis time period to proceed with rapid treatment and therapy which can limit late diagnosis of cervical cancer.

尽管公开可用的基准数据集可为新算法开发与结果对比提供基础，但在基于人工智能的自动化医学研究中，从真实临床场景采集的医院来源数据同样至关重要——这类数据可用于疾病诊断、按照标准流程对预测疾病开展组织学分期分类或任意类别识别，从而让开发出的算法在对应区域的临床场景中具备尽可能高的准确率。本数据集仓库支持与完整决策支持系统相关的图像分割及最终分类研究工作。液基细胞学（Liquid based cytology）是宫颈癌筛查手段之一。本仓库共包含963张图像，依据贝塞斯达系统（The Bethesda System）的标准，被划分为四组图像，分别对应宫颈癌的癌前与癌性病变四大类别。这些巴氏涂片图像由徕卡ICC50 HD显微镜以40倍放大倍率拍摄，采集自460名患者，样本采用液基细胞学技术制备。对细胞水平异常变化进行显微镜观察，可检测出恶性或癌前病变特征。该人工检测流程耗时较长，且易受观察者间及观察者内差异影响，因此计算机辅助诊断能够缩短整体疾病诊断周期，助力快速开展治疗与干预，从而减少宫颈癌晚期诊断的情况。

搜集汇总

背景与挑战

背景概述

该数据集是一个用于宫颈癌多类诊断的液基细胞学巴氏涂片图像集合，包含963张图像，根据The Bethesda System标准分为四个类别，代表癌前病变和癌性病变。图像来自460名患者，在40倍放大倍数下捕获，旨在支持AI研究中的图像分割和分类任务，以开发决策支持系统并提高诊断效率。