Anatomical correlations for a hierarchical multi-atlas segmentation of the pancreas in

CT images

Oscar A. Jiménez del Toro University of Applied Sciences Western Switzerland (HES-SO)

Overview •  Introduction •  VISCERAL •  Method

•  Multi-atlas segmentation •  Image registration •  Hierarchical registration approach

•  Pancreas segmentation •  Results

2

Overview •  Introduction •  VISCERAL •  Method

•  Multi-atlas segmentation •  Image registration •  Hierarchical registration approach

•  Pancreas segmentation •  Results

3

Introduction •  Anatomical segmentation is fundamental for

further image analysis1

•  Different methods proposed2,3 (regression random forests, level set…)

•  Comparison of multiple approaches for the same public dataset is uncommon

4

VISual Concept Extraction challenge in RAdioLogy

•  EU funded project (2012-2015) –  HES-SO, ETHZ, UHD, MUW, TUW,

Gencat •  Organize competitions on medical

image analysis on big data •  All computation done in the cloud •  Segmentation benchmark •  Retrieval benchmark

•  Annotation by medical doctors

Cloud environment

Benchmark 2 Anatomy •  Automatic segmentation of

anatomical structures (20) and landmark detection

•  Define challenges in large scale data (aprox. 10TB) processing

•  CT and MR images (contrast-enhanced and non-enhanced)

Overview •  Introduction •  VISCERAL •  Method

•  Multi-atlas segmentation •  Image registration •  Hierarchical registration approach

•  Pancreas segmentation •  Results

8

Overview •  Introduction •  VISCERAL •  Method

•  Multi-atlas segmentation •  Image registration •  Hierarchical registration approach

•  Pancreas segmentation •  Results

9

Hierarchical multi-atlas segmentation •  Use multiple atlases for

the estimation on a target image

•  Global and local alignment •  Hierarchical selection of

the registrations improves results

•  Label fusion

Image Registration •  Atlas = Patient volume + labels •  Coordinate transformation

that increases spatial correlation – Affine: Rotate, translate, scale – B-spline: Non-rigid

Right Kidney

Liver

Global alignment

Urinary Bladder Right Lung Left Lung 1st Lumbar Vertebra

Gall- bladder

Left Kidney Trachea

Spleen

2nd Local Affine

Hierarchical Registration approach

Affine

Local Affine

B-spline non-rigid

Label fusion •  Majority voting threshold

•  Classification on a per-voxel basis

•  Threshold optimization

Overview •  Introduction •  VISCERAL •  Method

•  Multi-atlas segmentation •  Image registration •  Hierarchical registration approach

•  Pancreas segmentation •  Results

14

Overview •  Introduction •  VISCERAL •  Method

•  Multi-atlas segmentation •  Image registration •  Hierarchical registration approach

•  Pancreas segmentation •  Results

15

Right Kidney

Liver

Global alignment

Urinary Bladder Right Lung Left Lung 1st Lumbar Vertebra

Gall- bladder

Left Kidney Trachea

Spleen

2nd Local Affine

Pancreas segmentation Affine

Local Affine

B-spline non-rigid

Right Kidney

Liver

Global alignment

Urinary Bladder Right Lung Left Lung 1st Lumbar Vertebra

Gall- bladder

Left Kidney Trachea

Spleen

2nd Local Affine

Affine

Local Affine

B-spline non-rigid

Liver

Right Kidney

Pancreas segmentation

Experimental setup •  VISCERAL Benchmark 1 testset •  10 contrast-enhanced CT volumes of the trunk •  Added to segmentation method of 10

structures: – Liver, lungs, kidneys, gallbladder, urinary bladder,

1st lumbar vertebra, trachea and spleen •  7 independent atlases as trainingset

Results •  Average DICE score for Pancreas: 0.52

Structure DICE Rank in VISCERAL Benchmark 1

Liver 0.918 1st Right Kidney 0.913 1st Left Kidney 0.921 1st Right Lung 0.965 3rd Left Lung 0.955 3rd Spleen 0.852 3rd Trachea 0.836 2nd Gallbladder 0.566 1st Urinary bladder 0.7 3rd 1st Lumbar vertebra 0.522 2nd

Results •  Average DICE score for Pancreas: 0.52

Structure DICE Rank in VISCERAL Benchmark 1

Liver 0.918 1st Right Kidney 0.913 1st Left Kidney 0.921 1st Right Lung 0.965 3rd Left Lung 0.955 3rd Spleen 0.852 3rd Trachea 0.836 2nd Gallbladder 0.566 1st Urinary bladder 0.7 3rd 1st Lumbar vertebra 0.522 2nd

Conclusion •  Full automatic method •  Requires little or no feedback from the user •  Showed robustness in the segmentation of

multiple structures with high overlap •  Fared well when compared to other methods

of the VISCERAL Benchmark 1 •  Future work:

–  Extend to method to other modalities (CTwb ISBI challenge, MR) –  Test in a bigger dataset for VISCERAL Benchmark 2 Anatomy

Sierre, Switzerland

Questions???