ISIC 2018 Leaderboards

Task 1: Lesion Boundary Segmentation Task 2: Lesion Attribute Detection Task 3: Lesion Diagnosis

Rank 66 total	Team 66 unique teams	Approach	Used External Data 8 yes	Primary Metric Value Jaccard Index
1	MT MT	MaskRcnn2+segmentation	No	0.802

2	Holidayburned Holidayburned	ensemble_with_CRF_v3	No	0.799

3	imsight imsight	Automatic Skin Lesion Segmentation by DCNN	No	0.799

4	Tencent Youtu Lab Tencent Youtu Lab	Skin Lesion Segmentation with Adversarial Learning	No	0.798

5	NMN_team NMN_team	segmentation_ensembleALL_Th0.80_Tl0.65	No	0.796

6	GPM-UC3M GPM-UC3M	SR FCN Init 2	No	0.788

7	Andrey Sorokin Andrey Sorokin	Mask-RCNN with SGD optimizer	No	0.779

8	DC DC	maskrcnn	No	0.777

9	Weill Cornell Medicine Weill Cornell Medicine	Deep Unet	Yes	0.773

10	Opsins Opsins	Transfer learning based CNN segmentation	No	0.771

11	DAISYLab DAISYLab	Ensemble of four models	Yes	0.769

12	SJTU_MJTC SJTU_MJTC	Deep attention-guided fusion network for lesion segmentation	No	0.765

13	Chongqing Institute of Green and Intelligent Technology, Chinese Academy of Scie Chongqing Institute of Green and Intelligent Technology, Chinese Academy of Scie	Combining FCNs, CRFs and Data Augmentation for Lesion Boundary Segmentation	No	0.761

14	Manu Goyal Manu Goyal	Test-D	No	0.760

15	Le-Health Le-Health	Mulit-Scale PSPNet	No	0.758

16	Li Li	Resnet50	No	0.758

17	BMCV Heidelberg BMCV Heidelberg	A two-step deep learning approach for skin lesion segmentation	No	0.757

18	Università degli Studi di Modena e Reggio Emilia (AImage Lab .zip) Università degli Studi di Modena e Reggio Emilia (AImage Lab .zip)	conv-deconv, bagging, data augmentation using GANs.	Yes	0.755

19	Tecnológico de Costa Rica - Comunidad ciencia de los datos Tecnológico de Costa Rica - Comunidad ciencia de los datos	U-Net with CRF post-processing	No	0.754

20	Ciklum Ciklum	Dilated U-net	No	0.753

21	LTS5 LTS5	Fully Convolutional Neural Network using dense block	Yes	0.747

22	CBML University of Piraeus CBML University of Piraeus	Mask R-CNN	No	0.746

23	Cleveland Clinic Cleveland Clinic	deep fully convolutional network with adam optimization	No	0.742

24	QuindiTech: : Xuan Li, Jizong Peng QuindiTech: : Xuan Li, Jizong Peng	Dilation Residual Net with ensemble (Xuan)	No	0.740

25	BIL, NTU BIL, NTU	Deep learning and image processing framework - ensemble 3	No	0.739

26	Nile University - MIIP Nile University - MIIP	Lesion Segmentation using Deep-Nevus Architecture	No	0.738

27	BioImaging-KHU BioImaging-KHU	Lesion_Segmentation	No	0.737

28	Jia Hua Ng Jia Hua Ng	DEXTR Simulated	No	0.736

29	BMIT-USYD BMIT-USYD	Segmentation using Adversarial Learning based Data Argumentation (Ensembled)	No	0.734

30	ECUST ECUST	Skin Lesion Segmentation based on Deeply Supervised Salient Object Detection	No	0.733

31	Ask Sina Ask Sina	Approach 2 : Min of Max prediction for all 4 Fold	No	0.731

32	MIA Group, Sun Yat-sen University MIA Group, Sun Yat-sen University	ensemble multi-scale U-Nets with postprocessing	No	0.730

33	WVU WVU	unet-gan-perceptual	Yes	0.729

34	IRCV IRCV	SLSDeep with EPE Loss	No	0.729

35	The Homeboy's The Homeboy's	Classifier approach	No	0.728

36	RECOD Titans RECOD Titans	Single Model - UNet Trained on Challenge Data Only	No	0.728

37	SAIIP-MIA SAIIP-MIA	Residual U-net	No	0.728

38	SSC-SJTU SSC-SJTU	Saliency-based Skin Lesion Segmentation by Deep Network	No	0.727

39	UNIST_BMIPL UNIST_BMIPL	Multiscale Approach for Lesion Segmentation	No	0.726

40	QuindiTech: Xuan Li, Jizong Peng QuindiTech: Xuan Li, Jizong Peng	111	No	0.726

41	Texas A&M Aggies Texas A&M Aggies	Course-to-Fine PSPNet	No	0.726

42	uestc_kb545 uestc_kb545	DL	No	0.725

43	LME LME	Multitask Framework for Skin Lesion Detection and Segmentation	No	0.724

44	Mammoth Mammoth	final_test_crf_0.753_upsampling	No	0.715

45	AI Toulouse AI Toulouse	Segmentation by GrassNet with rotationORflipORnoise	No	0.712

46	Virtual Expertise Virtual Expertise	Unet	No	0.712

47	BUPT Pris BUPT Pris	IVU: Improved VGG-Unet model in ISIC 2018	No	0.711

48	LABCIN LABCIN	Blue channel segmentation with Otsu Thresholding and preprocessing techniques	No	0.710

49	NTHU CVLab NTHU CVLab	Deep learning with GCN model	No	0.706

50	Tandon Titans Tandon Titans	U-net without post-processing	Yes	0.704

51	CDSLab CDSLab	SDI++	No	0.702

52	Redha Ali, Russell C. Hardie, Manawaduge Supun De Silva, and Temesguen Messay Ke Redha Ali, Russell C. Hardie, Manawaduge Supun De Silva, and Temesguen Messay Ke	Combining Deep and Handcrafted Image Features	No	0.698

53	Insight Centre for data analytics Insight Centre for data analytics	Deep Residual Architecture for Skin Lesion Segmentation	No	0.697

54	Q_PRML_laboratory Q_PRML_laboratory	Using deep encoder-decoder network for skin lesion segmentation	No	0.696

55	UnB UnB	Custom U-Net with Resnet34 backbone as encoder- Unet34 256x256 BestDice Strategy	No	0.693

56	PA_Tech PA_Tech	Effective u-net model structure and model integration strategy	Yes	0.689

57	medical image analysis lab - SFU medical image analysis lab - SFU	Starshape with dense upsampling	No	0.688

58	University of York University of York	Multitask learning, single CNN for all three tasks, segmentation via FCN	Yes	0.670

59	HUST_czj HUST_czj	fully convolutional networks	No	0.667

60	Department of Dermatology, University of Rzeszów, Poland Department of Dermatology, University of Rzeszów, Poland	FusionNet	No	0.666

61	University of Washington CSE University of Washington CSE	A Deep Fully Convolutional Network for Segmentation of Dermoscopic Images	No	0.665

62	University of Dayton, Signal and Image Processing Lab University of Dayton, Signal and Image Processing Lab	Traditional classifier with adaptive threshold guided by SVM regression	No	0.660

63	ITAKA ITAKA	U-Net variant	No	0.653

64	Signal and Image Processing Lab Signal and Image Processing Lab	Skin Lesion Segmentation using Clustering and Regression for ISIC 2018	No	0.618

65	LAPI, Image Processing and Analysis Laboratory (ETTI, Politehnica, Bucuresti, RO LAPI, Image Processing and Analysis Laboratory (ETTI, Politehnica, Bucuresti, RO	Ensemble of [Global Thresholds on RGB & Gray] and [K-Means clustering]	No	0.590

66	Math & Stat Dept., UNC-Greensboro, USA and CSIE Dept., NTNU, Taiwan Math & Stat Dept., UNC-Greensboro, USA and CSIE Dept., NTNU, Taiwan	Persistent homology based segmentation algorithm	No	0.532

Report an Issue

The MILK10k Challenge is open!

Join the competition!

ISIC 2018 Leaderboards