Brain CT Segmentation for medical diagnostics

The dataset contains over 1,000 studies encompassing 10 pathologies, providing a comprehensive resource for advancing research in brain imaging techniques. It comprises a wide variety of CT scans aimed at facilitating segmentation tasks related to brain tumors, lesions, and other brain structures. The data is provided in nii format and includes both volumetric data and the corresponding masks for each study, facilitating comprehensive… See the full description on the dataset page: https://huggingface.co/datasets/UniDataPro/brain-ct-segmentation.

This dataset, featured in the RSNA Intracranial Hemorrhage Detection challenge on Kaggle, offers a rich collection of brain CT images. It is meticulously categorized into seven distinct classes: 'none', 'epidural', 'intraparenchymal', 'intraventricular', 'subarachnoid', and 'subdural'. Each category is represented by 1000 DICOM files, providing a balanced and extensive dataset for analysis and machine learning applications. Dataset Details Categories: None, Epidural, Intraparenchymal, Intraventricular, Subarachnoid, Subdural Files: 1000 DICOM files per category Total Images: 6000 DICOM files Source: RSNA Intracranial Hemorrhage Detection Challenge on Kaggle (https://www.kaggle.com/competitions/rsna-intracranial-hemorrhage-detection/data) Use Case

This dataset contains over 9,000 head CT scans, each labeled as normal or abnormal. Each scan contains a reconstructed image (stored in our institution’s PACS and saved as DICOMs) and a corresponding sinogram (simulated via GE’s CatSim software and saved as numpy arrays). The reconstructed images are 512x512 pixels with a variable number of axial slices per scan. The sinograms are 984x888 pixels with a variable number of axial slices per scan. The full dataset is 1.3T. We retrospectively collected the head CT scans (acquired between 2001 – 2014) from our institution’s PACS, selected according to the following criteria: non-contrast CT of the head acquired in axial mode on a GE scanner and pixel spacing of 0.49 or 0.60 mm in the axial plane. The reading radiologist designated each CT scan as normal or abnormal at the time of original image interpretation; these designations were given as part of standard clinical procedure and not modified during dataset curation. We used GE’s CatSim, a validated simulation software for GE machines, to simulate high-fidelity sinograms of each head CT scan. If you use this dataset, please cite our paper (https://pubs.rsna.org/doi/abs/10.1148/ryai.2021200229). Additionally, part of this dataset was used in the RSNA Intracranial Hemorrhage Detection Challenge (https://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection). Labels for hemorrhage can be found in the Kaggle download.

The Magnetic Resonance - Computed Tomography (MR-CT) Jordan University Hospital (JUH) dataset has been collected after receiving Institutional Review Board (IRB) approval of the hospital and consent forms have been obtained from all patients. All procedures has been carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki).

The dataset consists of 2D image slices extracted using the RadiAnt DICOM viewer software. The extracted images are transformed to DICOM image data format with a resolution of 256x256 pixels. There are a total of 179 2D axial image slices referring to 20 patient volumes (90 MR and 89 CT 2D axial image slices). The dataset contains MR and CT brain tumour images with corresponding segmentation masks. The MR images of each patient were acquired with a 5.00mm T Siemens Verio 3T using a T2-weighted without contrast agent, 3 Fat sat pulses (FS), 2500-4000 TR, 20-30 TE, and 90/180 flip angle. The CT images were acquired with Siemens Somatom scanner with 2.46mGY.cm dose length, 130KV voltage, 113-327 mAs tube current, topogram acquisition protocol, 64 dual source, one projection, and slice thickness of 7.0mm. Smooth and sharp filters have been applied to the CT images. The MR scans have a resolution of 0.7x0.6x5 mm^3, while the CT scans have a resolution of 0.6x0.6x7 mm^3.

More information and the application of the dataset can be found in the following research paper:

Alaa Abu-Srhan; Israa Almallahi; Mohammad Abushariah; Waleed Mahafza; Omar S. Al-Kadi. Paired-Unpaired Unsupervised Attention Guided GAN with Transfer Learning for Bidirectional Brain MR-CT Synthesis. Comput. Biol. Med. 136, 2021. doi: https://doi.org/10.1016/j.compbiomed.2021.104763.

CQ500 dataset of 491 Computed tomography scans with 193,317 slices Anonymized dicoms for all the scans and the corresponding radiologists reads. ![]() Paper:

Brain Stroke CT Image Dataset

The dataset consists of CT brain scans with cancer, tumor, and aneurysm. Each scan represents a detailed image of a patient's brain taken using CT (Computed Tomography).

Many research applications of neuroimaging use magnetic resonance imaging (MRI). As such, recommendations for image analysis and standardized imaging pipelines exist. Clinical imaging, however, relies heavily on X-ray computed tomography (CT) scans for diagnosis and prognosis. Currently, there is only one image processing pipeline for head CT, which focuses mainly on head CT data with lesions. We present tools and a complete pipeline for processing CT data, focusing on open-source solutions, that focus on head CT but are applicable to most CT analyses. We describe going from raw DICOM data to a spatially normalized brain within CT presenting a full example with code. Overall, we recommend anonymizing data with Clinical Trials Processor, converting DICOM data to NIfTI using dcm2niix, using BET for brain extraction, and registration using a publicly-available CT template for analysis.

DATASET STRUCTURE

The dataset can be downloaded from https://doi.org/10.5281/zenodo.7260705 and a detailed description is offered at "synthRAD2023_dataset_description.pdf".

The training datasets for Task1 is in Task1.zip, while for Task2 in Task2.zip. After unzipping, each Task is organized according to the following folder structure:

Task1.zip/

├── Task1

├── brain

├── 1Bxxxx

├── mr.nii.gz

├── ct.nii.gz

└── mask.nii.gz

├── ...

└── overview

├── 1_brain_train.xlsx

├── 1Bxxxx_train.png

└── ...

└── pelvis

├── 1Pxxxx

├── mr.nii.gz

├── ct.nii.gz

├── mask.nii.gz

├── ...

└── overview

├── 1_pelvis_train.xlsx

├── 1Pxxxx_train.png

└── ....

Task2.zip/

├──Task2

├── brain

├── 2Bxxxx

├── cbct.nii.gz

├── ct.nii.gz

└── mask.nii.gz

├── ...

└── overview

├── 2_brain_train.xlsx

├── 2Bxxxx_train.png

└── ...

└── pelvis

├── 2Pxxxx

├── cbct.nii.gz

├── ct.nii.gz

├── mask.nii.gz

├── ...

└── overview

├── 2_pelvis_train.xlsx

├── 2Pxxxx_train.png

└── ....

Each patient folder has a unique name that contains information about the task, anatomy, center and a patient ID. The naming follows the convention below:

[Task] [Anatomy] [Center] [PatientID]

1 B A 001

In each patient folder, three files can be found:

• ct.nii.gz: CT image

• mr.nii.gz or cbct.nii.gz (depending on the task): CBCT/MR image

• mask.nii.gz:image containing a binary mask of the dilated patient outline

For each task and anatomy, an overview folder is provided which contains the following files:

• [task]_[anatomy]_train.xlsx: This file contains information about the image acquisition protocol for each patient.

• [task][anatomy][center][PatientID]_train.png: For each patient a png showing axial, coronal and sagittal slices of CBCT/MR, CT, mask and the difference between CBCT/MR and CT is provided. These images are meant to provide a quick visual overview of the data.

DATASET DESCRIPTION

This challenge dataset contains imaging data of patients who underwent radiotherapy in the brain or pelvis region. Overall, the population is predominantly adult and no gender restrictions were considered during data collection. For Task 1, the inclusion criteria were the acquisition of a CT and MRI during treatment planning while for task 2, acquisitions of a CT and CBCT, used for patient positioning, were required. Datasets for task 1 and 2 do not necessarily contain the same patients, given the different image acquisitions for the different tasks.

Data was collected at 3 Dutch university medical centers:

• Radboud University Medical Center

• University Medical Center Utrecht

• University Medical Center Groningen

For anonymization purposes, from here on, institution names are substituted with A, B and C, without specifying which institute each letter refers to.

The following number of patients is available in the training set.

Training

	Brain	Pelvis
	Center A	Center B	Center C	Total	Center A	Center B	Center C	Total
Task 1	60	60	60	180	120	0	60	180
Task 2	60	60	60	180	60	60	60	180

Each subset generally contains equal amounts of patients from each center, except for task 1 brain, where center B had no MR scans available. To compensate for this, center A provided twice the number of patients than in other subsets.

Validation

	Brain	Pelvis
	Center A	Center B	Center C	Total	Center A	Center B	Center C	Total
Task 1	10	10	10	30	20	0	10	30
Task 2	10	10	10	30	10	10	10	30

Testing

	Brain	Pelvis
	Center A	Center B	Center C	Total	Center A	Center B	Center C	Total
Task 1	20	20	20	60	40	0	20	60
Task 2	20	20	20	60	20	20	20	60

In total, for all tasks and anatomies combined, 1080 image pairs (720 training, 120 validation, 240 testing) are available in this dataset. This repository only contains the training data.

All images were acquired with the clinically used scanners and imaging protocols of the respective centers and reflect typical images found in clinical routine. As a result, imaging protocols and scanner can vary between patients. A detailed description of the imaging protocol for each image, can be found in spreadsheets that are part of the dataset release (see dataset structure).

Data was acquired with the following scanners:

• Center A:

  • MRI: Philips Ingenia 1.5T/3.0T

  • CT: Philips Brilliance Big Bore or Siemens Biograph20 PET-CT

  • CBCT: Elekta XVI

• Center B:

  • MRI: Siemens MAGNETOM Aera 1.5T or MAGNETOM Avanto_fit 1.5T

  • CT: Siemens SOMATOM Definition AS

  • CBCT: IBA Proteus+ or Elekta XVI

• Center C:

  • MRI: Siemens Avanto fit 1.5T or Siemens MAGNETOM Vida fit 3.0T

  • CT: Philips Brilliance Big Bore

  • CBCT: Elekta XVI

For task 1, MRIs were acquired with a T1-weighted gradient echo or an inversion prepared - turbo field echo (TFE) sequence and collected along with the corresponding planning CTs for all subjects. The exact acquisition parameters vary between patients and centers. For centers B and C, selected MRIs were acquired with Gadolinium contrast, while the selected MRIs of center A were acquired without contrast.

For task 2, the CBCTs used for image-guided radiotherapy ensuring accurate patient position were selected for all subjects along with the corresponding

This dataset comprises non-contrast CT scans of patients admitted with subarachnoid hemorrhage. All scans are provided in the NIfTI format. Details of the image preprocessing, which encompasses anonymization, resampling, skull stripping, and intensity normalization, are available at https://github.com/smcch/Subarachnoid_Hemorrhage_segmentation_and_mortality_prediction.

Each scan is accompanied by a corresponding NIfTI volume, showcasing the expert manual segmentation of the hemorrhage.

Ovewview

This is 8 folds se_resnext101_32x4d checkpoints trained on brain CT datasets provided by RSNA. Because kaggle datasets did not allow the size of dataset exceeding 6 GB, I split the dataset into to two and this is the first part of the dataset.

You can download the training dataset here. https://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection/data

Description

The detail of this model can be found here. https://github.com/appian42/kaggle-rsna-intracranial-hemorrhage

These checkpoints are generated by running this script.

Acknowledgements

RSNA https://www.rsna.org/

CT Scans of Brain for computer vision tasks

Dataset comprises over 70,000 studies, including 20,000+ studies with protocols developed by medical professionals and 50,000+ studies without protocols. It is designed to facilitate research in the field of medical imaging, specifically focusing on the detection and analysis of brain pathologies, including 5 distinct pathologies such as brain tumors, brain hemorrhages, and brain cancers. By utilizing this dataset, researchers can… See the full description on the dataset page: https://huggingface.co/datasets/UniDataPro/ct-scans-of-brain.

Dataset of CT scans of the brain includes over 1,000 studies that highlight various pathologies such as acute ischemia, chronic ischemia, tumor, and etc

Dataset of brain CT scans featuring 70,000+ studies with medical-grade protocols for healthcare AI and diagnostic research

Ovewview

This is 8 folds se_resnext101_32x4d checkpoints trained on brain CT datasets provided by RSNA. Because kaggle datasets did not allow the size of dataset exceeding 6 GB, I split the dataset into to two and this is the latter part of the dataset.

You can download the training dataset here. https://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection/data

Description

The detail of this model can be found here. https://github.com/appian42/kaggle-rsna-intracranial-hemorrhage

These checkpoints are generated by running this script.

Acknowledgements

RSNA https://www.rsna.org/

Ovewview

This is 8 folds se_resnext50_32x4d checkpoints trained on brain CT datasets provided by RSNA.

You can download the training dataset here. https://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection/data

Description

The detail of this model can be found here. https://github.com/appian42/kaggle-rsna-intracranial-hemorrhage

These checkpoints are generated by running this script.

Acknowledgements

RSNA https://www.rsna.org/

Ct Brain Hemorrhage

## Overview

Ct Brain Hemorrhage is a dataset for object detection tasks - it contains Cxr Lesion3 annotations for 3,568 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

RIDER Neuro MRI contains imaging data on 19 patients with recurrent glioblastoma who underwent repeat imaging sets. These images were obtained approximately 2 days apart (with the exception of one patient, RIDER Neuro MRI-1086100996, whose images were obtained one day apart).

DCE‐MRI: All 19 patients had repeat dynamic contrast‐enhanced MRI (DCE‐MRI) datasets on the same 1.5T imaging magnet. On the basis of T2‐weighted images, technologists chose 16 image locations using 5mm thick contiguous slices for the imaging. For T1 mapping, multi‐flip 3D FLASH images were obtained using flip angles of 5, 10, 15, 20, 25 and 30 degrees, TR of 4.43 ms, TE of 2.1 ms, 2 signal averages. Dynamic images were obtained during the intravenous injection of 0.1mmol/kg of Magnevist intravenous at 3ccs/second, started 24 seconds after the scan had begun. The dynamic images were acquired using a 3D FLASH technique, using a flip angle of 25 degrees, TR of 3.8 ms, TE of 1.8 ms using a 1 x1 x 5mm voxel size. The 16 slice imaging set was obtained every 4.8 sec.

DTI: Seventeen of the 19 patients also obtained repeat diffusion tensor imaging (DTI) sets. Whole brain DTI were obtained using TR 6000ms, TE 100 ms, 90 degree flip angle, 4 signal averages, matrix 128 x 128, 1.72 x 1.72 x 5 mm voxel size, 12 tensor directions, iPAT 2, b value of 1000 sec/mm2 .

Contrast‐enhanced 3D FLASH: All 19 patients underwent whole brain 3D FLASH imaging in the sagittal plane after the administration of Magnevist. For this sequence, TR was 8.6 ms, TE 4.1 ms, 20 degree flip angle, 1 signal average, matrix 256 x 256; 1mm isotropic voxel size.

Contrast‐enhanced 3D FLAIR: All 17 patients who had repeat DTI sets also had 3D FLAIR sequences in the sagittal plane after the administration of Magnevist. For this sequence, the TR was 6000 ms, TE 353 ms, and TI 2200ms; 180 degree flip angle, 1 signal average, matrix 256 x 216; 1 mm isotropic voxel size. Note: before transmission to NCIA, all image sets with 1mm isotropic voxel size were “defaced” using MIPAV software or manually.

About the RIDER project

The Reference Image Database to Evaluate Therapy Response (RIDER) is a targeted data collection used to generate an initial consensus on how to harmonize data collection and analysis for quantitative imaging methods applied to measure the response to drug or radiation therapy. The National Cancer Institute (NCI) has exercised a series of contracts with specific academic sites for collection of repeat "coffee break," longitudinal phantom, and patient data for a range of imaging modalities (currently computed tomography [CT] positron emission tomography [PET] CT, dynamic contrast-enhanced magnetic resonance imaging [DCE MRI], diffusion-weighted [DW] MRI) and organ sites (currently lung, breast, and neuro). The methods for data collection, analysis, and results are described in the new Combined RIDER White Paper Report (Sept 2008):

• RIDER White Paper: Combined contracts report ( Sept 2008) PDF

The long term goal is to provide a resource to permit harmonized methods for data collection and analysis across different commercial imaging platforms to support multi-site clinical trials, using imaging as a biomarker for therapy response. Thus, the database should permit an objective comparison of methods for data collection and analysis as a national and international resource as described in the first RIDER white paper report (2006):

• RIDER White Paper: Executive Summary PDF
• RIDER White Paper: Editorial in Nature.com

The dataset can be downloaded from https://doi.org/10.5281/zenodo.7868169 and a detailed description is offered at https://doi.org/10.5281/zenodo.7260704 in the "synthRAD2023_dataset_description.pdf".

The input of the validation datasets for Task1 is in Task1_val.zip, while for Task2 in Task2_val.zip. After unzipping, each Task is organized according to the following folder structure:

Task1_val.zip/

├── Task1

├── brain

├── 1Bxxxx


  ├── mr.nii.gz


  └── mask.nii.gz


├── ...

└── overview

  ├── 1_brain_val.xlsx


  ├── 1Bxxxx_val.png


 └── ...  

└── pelvis

├── 1Pxxxx


  ├── mr.nii.gz


  ├── mask.nii.gz

├── ...

└── overview

  ├── 1_pelvis_val.xlsx


  ├── 1Pxxxx_val.png


  └── ....

Task2_val.zip/

├──Task2

├── brain

├── 2Bxxxx


  ├── cbct.nii.gz


  └── mask.nii.gz


├── ...

└── overview

  ├── 2_brain_val.xlsx


  ├── 2Bxxxx_val.png


 └── ...  

└── pelvis

├── 2Pxxxx


  ├── cbct.nii.gz


  ├── mask.nii.gz

├── ...

└── overview

 ├── 2_pelvis_val.xlsx


 ├── 2Pxxxx_val.png


 └── ....

Each patient folder has a unique name that contains information about the task, anatomy, center and a patient ID. The naming follows the convention below:

[Task] [Anatomy] [Center] [PatientID]

1 B A 001

In each patient folder, two files can be found:

mr.nii.gz or cbct.nii.gz (depending on the task): CBCT/MR image

mask.nii.gz: image containing a binary mask of the dilated patient outline

For each task and anatomy, an overview folder is provided which contains the following files:

[task]_[anatomy]_val.xlsx: This file contains information about the image acquisition protocol for each patient.

[task][anatomy][center][PatientID]_val.png: For each patient a png showing axial, coronal and sagittal slices of CBCT/MR, CT, mask and the difference between CBCT/MR and CT is provided. These images are meant to provide a quick visual overview of the data.

DATASET DESCRIPTION

This challenge dataset contains imaging data of patients who underwent radiotherapy in the brain or pelvis region. Overall, the population is predominantly adult and no gender restrictions were considered during data collection. For Task 1, the inclusion criteria were the acquisition of a CT and MRI during treatment planning while for task 2, acquisitions of a CT and CBCT, used for patient positioning, were required. Datasets for task 1 and 2 do not necessarily contain the same patients, given the different image acquisitions for the different tasks.

Data was collected at 3 Dutch university medical centers:

Radboud University Medical Center;

University Medical Center Utrecht;

University Medical Center Groningen.

For anonymization purposes, from here on, institution names are substituted with A, B and C, without specifying which institute each letter refers to.

The following number of patients is available in the validation set.

Validation

Brain

Pelvis

Center A

Center B

Center C

Total

Center A

Center B

Center C

Total

Task 1

10

10

10

30

20

0

10

30

Task 2

10

10

10

30

10

10

10

30

In total, for all tasks and anatomies combined, 120 image pairs are available in this dataset. This repository only contains the validation data. The training data is provided at: https://doi.org/10.5281/zenodo.7260704.

All images were acquired with the clinically used scanners and imaging protocols of the respective centers and reflect typical images found in clinical routine. As a result, imaging protocols and scanner can vary between patients. A detailed description of the imaging protocol for each image, can be found in spreadsheets that are part of the dataset release (see dataset structure).

Data was acquired with the following scanners:

Center A:

MRI: Philips Ingenia 1.5T/3.0T

CT: Philips Brilliance Big Bore or Siemens Biograph20 PET-CT

CBCT: Elekta XVI

Center B:

MRI: Siemens MAGNETOM Aera 1.5T or MAGNETOM Avanto_fit 1.5T

CT: Siemens SOMATOM Definition AS

CBCT: IBA Proteus+ or Elekta XVI

Center C:

MRI: Siemens Avanto fit 1.5T or Siemens MAGNETOM Vida fit 3.0T

CT: Philips Brilliance Big Bore

CBCT: Elekta XVI

For task 1, MRIs were acquired with a T1-weighted gradient echo or an inversion prepared - turbo field echo (TFE) sequence and collected along with the corresponding planning CTs for all subjects. The exact acquisition parameters vary between patients and centers. For centers B and C, selected MRIs were acquired with Gadolinium contrast, while the selected MRIs of center A were acquired without contrast.

For task 2, the CBCTs used for image-guided radiotherapy ensuring accurate patient position were selected for all subjects along with the corresponding planning CT.

The following pre-processing steps were performed on the data:

Conversion from dicom to compressed nifti (nii.gz)

Rigid registration between CT and MR/CBCT

Anonymization (face removal, only for brain patients)

Patient outline segmentation (provided as a binary mask)

Crop MR/CBCT, CT and mask to remove background and reduce file sizes

The code used to preprocess the images can be found at: https://github.com/SynthRAD2023/. Detailed information about the dataset are provided in SynthRAD2023_dataset_description.pdf published here along with the data and will also be submitted to Medical Physics.

ETHICAL APPROVAL

Each institution received ethical approval from their internal review board/Medical Ethical committee:

UMC Utrecht approved not-WMO on 4/03/2022 with number 22/474 entitled: “Synthetizing computed tomography for radiotherapy Grand Challenge (SynthRAD)”.

UMC Groningen approved not-WMO on 20/07/2022 with number 202200310 entitled: “Synthesizing computed tomography for radiotherapy - Grand Challenge”.

Radboud UMC declared the study not-WMO on 17/10/2022 with number 2022-15950 entitled “Synthetizing computed tomography for radiotherapy Grand Challenge”.

CHALLENGE DESIGN

The overall challenge design can be found at https://doi.org/10.5281/zenodo.7746019.

Evaluation of the models trained on the IXI dataset.