Actinic Keratosis in dermatoscopy and High-Frequency Ultrasound

Data supporting work: "Automatic Assessment of AK Stage Based on Dermatoscopic and HFUS Imaging—A Preliminary Study" DOI: https://doi.org/10.3390/jcm13247499 and "Actinic keratosis staging in multimodal image data" DOI: 10.1016/j.cmpb.2026.109364 .

This dataset consist of 222 HFUS and dermatoscopic images registered from 74 patients with actinic keratosis (AK) aged 41-89 years. All experiments were performed following the protocol approved by the Bioethics Committee at the Poznan University of Medical Sciences, under reference number 523/23 on 29 June 2023. This study was conducted at the Department of Dermatology of Poznan University of Medical Sciences from June 2023 to April 2024. For each marker (AK site) dermatoscopy picture were taken (iPhone 8 camera coupled to DermLite DL5, 10× magnification) and then high-frequency images (DermaScan C, a Cortex Technology, Aalborg, Denmark device, linear 20 MHz probe) were taken as well.

For each lesion, the clinico-dermatoscopic classification of each AK based on a three-point Zalaudek scale was performed by 2 experienced dermatologists. In this dataset, 122 markers were classified as AK1, 51 as AK2 and 34 as AK3. The rest was photographs of clinically healthy skin (stage 0). Each patient is codes as PXXX, and each corresponding markers for patient are coded as PXXX_mX. Dermatosopy and HFUS images have corresponding names in folders HFUS and Dermatoscopic. Patient info (age and sex are presented in file patientID_age_sex.csv, where 0:Female, 1:Male). File markersID_staging.csv contains stage corresponding to each markers (where 1:AK1, 2:AK2, 3:AK3, 0:clinically healthy skin).

This research was supported by National Science Centre, Poland, research project No. 2023/07/X/ST6/00861 “The use of artificial intelligence methods in the diagnosis of actinic keratosis based on multimodal image data” and No. 2023/07/X/NZ5/00867 “The importance of high-frequency ultrasound in the assessment of actinic keratosis and the cancerization field”.

In file markersID_staging..csv there is a column "VERSION" added - value 1 marks patiens from the publication "Automatic Assessment of AK Stage Based on Dermatoscopic and HFUS Imaging—A Preliminary Study" and number 2 - added patients. This repository contains also data recorded from younger people - below 41 (see age column).

本研究的支撑数据关联两篇已发表论文：《基于皮肤镜与高频超声成像的AK分期评估——初步研究》（DOI: https://doi.org/10.3390/jcm13247499）与《多模态影像数据中的光化性角化病分期》（DOI: 10.1016/j.cmpb.2026.109364）。
本数据集包含74名41~89岁光化性角化病（Actinic Keratosis，AK）患者的222幅高频超声（High-Frequency Ultrasound，HFUS）与皮肤镜影像。本研究所有实验均遵循波兰波兹南医科大学生物伦理委员会于2023年6月29日批准的研究方案（审批编号：523/23）执行，研究周期为2023年6月至2024年4月，实施地点为波兹南医科大学皮肤病学系。
针对每处病灶标记点（AK病变部位），首先使用搭载DermLite DL5镜头（放大倍数10×）的iPhone 8相机采集皮肤镜图像，随后使用丹麦奥尔堡Cortex Technology公司生产的DermaScan C设备（配备20MHz线性探头）采集高频超声影像。由2名经验丰富的皮肤科医生基于三点Zalaudek量表，对每处病变的临床-皮肤镜特征进行AK分期分类。本数据集中共122处标记点被归类为AK1期、51处为AK2期、34处为AK3期，剩余标记点对应临床健康皮肤（0期）。
所有患者均以PXXX格式进行编码，每位患者对应的病灶标记点则以PXXX_mX格式编码。皮肤镜与HFUS影像分别存储于Dermatoscopic与HFUS文件夹中，文件名一一对应。患者的年龄与性别信息存储于patientID_age_sex.csv文件中，其中性别编码规则为：0代表女性，1代表男性。markersID_staging.csv文件包含每处标记点对应的分期信息，编码规则为：1对应AK1期、2对应AK2期、3对应AK3期、0对应临床健康皮肤。
本研究得到波兰国家科学中心两项研究项目的资助，项目编号分别为2023/07/X/ST6/00861《基于多模态影像数据的人工智能方法在光化性角化病诊断中的应用》与2023/07/X/NZ5/00867《高频超声在光化性角化病及癌化场评估中的价值》。
在markersID_staging.csv文件中新增了VERSION列：数值1对应《基于皮肤镜与高频超声成像的AK分期评估——初步研究》一文中的患者，数值2对应新增纳入的患者。本数据集仓库还收录了41岁以下受试者的影像数据，相关信息详见age列。