Deep Learning-based Dot and Globule Segmentation with Pixel and Blob-based Metrics for Evaluation -- Data


Deep learning (DL) applied to whole dermoscopic images has shown unprecedented accuracy in differentiating images of melanoma from benign lesions. We hypothesize that accuracy in whole-image deep learning suffers because whole lesion analysis lacks evaluation of dermoscopic structures. DL also suffers a “black box” characterization because it offers only probabilities to the physician and no visible structures. We propose the detection of structures called globules and dots as a means to improve precision in melanoma detection. We compare two encoder-decoder architectures to detect globule and dots: UNET vs. UNET++. For each of these architectures, we compare three pipelines: with test time augmentation (TTA), without TTA, and without TTA but with checkpoint ensembles. We use an SE-RESNEXT encoder and a symmetric decoder. The pixel-based F-1 scores for globule and dot detection based on UNET++ and UNET techniques with checkpoint ensembles were found to be 0.632 and 0.628, respectively. The blob-based UNET++ and UNET F-1 (50 percent intersection) scores were 0.696 and 0.685, respectively. This agreement score is over twice the statistical correlation score measured among specialists. We propose UNET++ globule and dot detection as a technique that offers two potential advantages: increased diagnostic accuracy and visible structure detection to better explain DL results. We present a public globule and dot database to aid progress in automatic detection of these structures.

Viewing Instructions

The datasets contain the three zip files with the following descriptions,

"" : jpeg images of the ISIC19 images for which globule masks were created

"globule" : pixel exact globule masks

"globule" : exact globule masks dilated using an ellipse morphological kernel with dimensions 3x3 (Used in the paper).

You may wish to download and view individual files.


Electrical and Computer Engineering


Suggested citations (for ISIC19):

  1. N. C. F. Codella et al., “Skin lesion analysis toward melanoma detection: A challenge at the 2017 International symposium on biomedical imaging (ISBI), hosted by the international skin imaging collaboration (ISIC),” in Proceedings - International Symposium on Biomedical Imaging, 2018, vol. 2018-April, pp. 168–172. doi: 10.1109/ISBI.2018.8363547.
  2. N. Codella et al., “Skin Lesion Analysis Toward Melanoma Detection 2018: A Challenge Hosted by the International Skin Imaging Collaboration (ISIC).” 2019.
  3. M. Combalia et al., “BCN20000: Dermoscopic Lesions in the Wild.” 2019.
  4. P. Tschandl, C. Rosendahl, and H. Kittler, “The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions,” Sci Data, vol. 5, p. 180161, 2018.

Keywords and Phrases

Machine Learning; Deep Learning; Data processing; Melanoma; Globules; Feature Segmentation

Document Type


Document Version

Final Version

File Format





© 2022 Missouri University of Science and Technology, All rights reserved.

Publication Date

May 2022