Cancer Imaging Phenomics Toolkit (CaPTk)...[1] Davatzikos et al., Cancer imaging phenomics toolkit:...
Transcript of Cancer Imaging Phenomics Toolkit (CaPTk)...[1] Davatzikos et al., Cancer imaging phenomics toolkit:...
GLISTRboost[11]
T1 T1-Gd T2 T2-FLAIR
EnhancingNon-enhancingEdema/Invasion
White MatterGray MatterCerebrospinal Fluid
Tumor labels
Healthy brain labels
Target Audience:
– Genera l Purpose Tools –
· Functionality ·
Primary Aim: To enable swift and efficient translation of cutting-edge academic research into clinically useful tools[1].
– Spec ia l i zed Appl icat ions –
Imaging Signature of EGFRvIII in GBM [5] Computational Study
of Brain Connectivity [6,7]
DTI-based
Resection Margin Estimation
Peritumoral Effects
of Glioblastoma
· Future Work ·
Interaction
Coordinate definition on
various tissue types Region AnnotationApproximation by a
sphere
References
[1] Davatzikos et al., Cancer imaging phenomics toolkit: quantitative imaging analytics for precision diagnostics and predictive modeling of clinical outcome,
Journal of Medical Imaging, 2018
[2] Shinohara et al., Statistical normalization techniques for magnetic resonance imaging, Neuroimage Clinical, 2014
[3] Gaonkar et al., Adaptive geodesic transform for segmentation of vertebrae on CT images, Medical Imaging, 2014
[4] Yushkevich et al., ITK-SNAP: An interactive tool for semi-automatic segmentation of multi-modality biomedical images, IEEE Eng Med Biol Soc., 2016
[5] Bakas et al., In vivo detection of EGFRvIII in glioblastoma via perfusion magnetic resonance imaging signature consistent with deep peritumoral infiltration,
Clinical Cancer Research, 2017
[6] Tunc et al., Individualized Map of white matter pathways: connectivity-based paradigm for neurosurgical planning, Neurosurgery, 2016
[7] Tunc et al., Automated tract extraction via atlas based Adaptive Clustering, NeuroImage, 2014
[8] Keller et al., Estimation of breast percent density in raw and processed full field digital mammography images, Medical Physics, 2012
[9] Keller et al., Preliminary evaluation of the publicly available Laboratory for Breast Radiodensity Assessment (LIBRA) software tool, Breast Cancer Research 2015
[10] Li et al., Predicting treatment response and survival of early-stage non-small cell lung cancer patients treated with stereotactic body radiation therapy using
unsupervised two-way clustering of radiomic features, Int. Workshop on Pulmonary Imaging, 2017
[11] Bakas et al., GLISTRboost: Combining multimodal MRI segmentation, registration, and biophysical tumor growth modeling with gradient boosting machines for
glioma segmentation, Springer, LNCS, 2016
[12] Bakas et al., Advancing TCGA glioma MRI collections with expert segmentation labels and radiomic features, Nature Scientific Data, 2017.
[13] Akbari et al., Imaging Surrogates of infiltration obtained via multiparametric imaging pattern analysis predict subsequent location of recurrence of glioblastoma,
Neurosurgery, 2016
[14] Rathore et al., Radiomic signature of infiltration in peritumoral edema predicts subsequent recurrence in glioblastoma: Implications for personalized
radiotherapy planning, Journal of Medical Imaging, 2018
[15] Macyszyn et al., Imaging patterns predict patient survival and molecular subtype in glioblastoma via machine learning techniques, Neuro-Oncology, 2016
[16] Rathore et al., Imaging pattern analysis reveals three distinct phenotypic subtypes of GBM with different survival rates, Neuro-Oncology, 2016
[17] Kamnitsas et al., Efficient multi-Scale 3D CNN with fully connected CRF for accurate brain lesion segmentation, Medical Image Analysis, 2016
Breast Density Assessment [8,9]
(Cancer Risk Estimation)
Applied to more than
50,000 mammography
screening exams
GERawPD: 7.54%
GEProcessedPD: 75.6%
HologicRawPD: 22.5%
HologicProcessedPD: 22.8%
Cancer Imaging Phenomics Toolkit (CaPTk): A Radio(geno)mics Platform for Quantitative Imaging Analytics on Computational Oncology
C. Davatzikos, D. Kontos, P. Yushkevich, R. Shinohara, Y. Fan, R. VermaCenter for Biomedical Image Computing and Analytics (CBICA), University of Pennsylvania
www.cbica.upenn.edu/captk
FUNDING: ITCR U24CA189523
Specialized Segmentation
CTPET
Lung Tumor [10] LIBRA [8,9]
Segmentation [3,4]
ITK-SNAPQuantitative Feature Extraction
Textural
Voxel-based/ROI based
texture features
Gray-Level
Matrices
Local Binary
Patterns
1st Order Statistics Histogram-related
Intensity-basedVolume / Morphology
Model Training Functionality | Brain: Distinct radiographic subtypes of GBM[14] | Breast: Comprehensive parenchymal texture characterization | Deep Learning pipelines for segmentation[17]
• Racial disparities
• Strong associations with breast cancer risk
• Density changes after bariatric surgery
@CBICAannounce
github.com/cbica
@CBICA
Published in Nature Scientific Data[12]
enriching the TCGA-GBM & TCGA-LGG datasets with
manual tumor segmentations and radiomic features,
publicly available on the TCIA webpage.
Personalized Radiation Dose Escalation in areas of
higher likelihood of recurrence: Application of our
predictive maps to a trial funded by Abramson Cancer
Center and to an NRG trial.
GBM Recurrence Prediction [13, 14]
(Predictive Maps of Peritumoral Infiltration)
High Probabilityof Recurrence
Low Probabilityof Recurrence
Post-recurrence scan, with nodular enhancement in the predicted areas
Infiltration heatmap in pre-op T1-Gd
Cross-Platform ExtendableOpen-Source
1. Clinical experts: facilitating use of complex algorithms for clinically relevant studies through a user-friendly interface.
2. Computational experts: allowing for batch-processing and integration of new algorithms.
Easy to use
Input MRI DenoisingBias
CorrectionHistogram
Matching
T1-G
dT2-F
LA
IR
Pre-processing [2]
GBM Survival Prediction [15]
Kaplan-Meier Survival Curves
Survival (Months)
Cu
mu
lati
ve s
urv
ival ra
te (
%)
HR(low&high): 10.64 (95% CI 5.9-19.3, p<0.001)
HR(med&high): 3.88 (95% CI 2.3-606, p<0.001)
HR(low&med): 2.77 (95% CI 1.8-4.2, p<0.001)
Low SPI
Medium SPI
High SPI
Web-access ible
ipp.cbica.upenn.edu
(HPC-shared resources for computationally
demanding pipelines)
Spatial Distribution Atlases
Cla
ss 1
Cla
ss 2
100
80
60
40
20
0
0 20 40 60 80