BlaVeS: Bladder Vessel Segmentation

<p> The <strong>Bladder Vessel Segmentation (BlaVeS)</strong> dataset is the first publicly available dataset for blood vessel segmentation in urinary bladder endoscopy images. It provides hand-labeled pixel-wise annotations of prominent blood vessels to support bladder navigation, tumor monitoring, and biopsy site localization. The dataset is designed to aid in developing deep learning models for improving orientation and navigation during cystoscopic procedures. </p> <p> A total of <strong>50 representative images</strong> were selected from four clinical cystoscopic video recordings (8 minutes in total). Each image has a <strong>512×512 pixel resolution</strong> and is stored in PNG format. </p> <h3>Dataset Structure</h3> <ul> <li><strong>pretraining/</strong> (For initial model training)</li> <li><strong>training/</strong> (Labeled images for training segmentation models)</li> <li><strong>training_augmented/</strong> (Extended training set with augmented samples)</li> <li> <strong>testing/</strong> <ul> <li><strong>images/</strong> (Input cystoscopic images) <ul> <li>input_image_1.png</li> <li>input_image_2.png</li> <li>...</li> </ul> </li> <li><strong>groundtruth/</strong> (Pixel-wise blood vessel masks) <ul> <li>mask_1.png</li> <li>mask_2.png</li> <li>...</li> </ul> </li> </ul> </li> </ul>