The state-of-the-art 3D anisotropic intracranial hemorrhage segmentation on non-contrast head CT: The INSTANCE challenge

Li, Xiangyu; Luo, Gongning; Wang, Kuanquan; Wang, Hongyu; Li, Jun; Liang, Xiao; Jiang, Jie; Song, Zhenghao; Zheng, Chunyue; Chi, Haokai; Xu, Manman; He, Yingte; Ma, Xinghua; Guo, Jingwen; Liu, Yifan; Chuanpu, Li,; Chen, Zeli; Siddiquee, Md Mahfuzur Rahman; Myronenko, Andriy; Sanner, Antoine; Mukhopadhyay, Anirban; Othman, Ahmed E.; Zhao, Xinqing; Liu, Weiping; Zhang, Jinhuang; Ma, Xiangyuan; Liu, Qinghui; MacIntosh, Bradley J.; Liang, Wei; Mazher, Moona; Qayyum, Abdul; Abramova, Valeriia; Lladó, Xavier; Li, Shuo

doi:10.48550/arxiv.2301.03281

Cited by 3 publications

(4 citation statements)

References 38 publications

(65 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Mask2Former [4] built on top of end-to-end object detection model [14] has made different segmentation tasks like instance, semantic and panoptic segmentation to be easily trained with a Transformer based architecture. However, the unitary end-to-end model is usually proposed in brain imaging problem such as the ICH challenge [15]. Since there are other segmentation tasks in CT head scan, our proposed transformer-based model simplified the architecture to perform multiple segmentation tasks.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Sequential Modelling for Medical Image Segmentation

How,

Rajapakse

2023

Preprint

View full text Add to dashboard Cite

An image can be seen as a long sequence of tokens with spatial structure hence an image segmentation task can be treated as a sequence-to-sequence task. Existing attention-based segmentation models have incorporated an up-sampling module at the pixel decoder, such design assumed the backbone to have multiple feature scales. Furthermore, the existing models did not emphasize the ability of sequential modelling for multi-tasking. In this paper, we propose to model image segmentation task as a long sequence prediction task with an improvisation on MegaByte. As MegaByte efficiently reduced the self-attention cost using multi-scale structure known as global and local models, we train two MegaByte models as an encoder and decoder for multi-tasking. Our multi-task model allows different segmentation tasks to be trained simultaneously and inferenced with a task prompt. The encoder takes in the image sequence to encode them as a key while the decoder takes in the conditional input as a query. As decoder conditional input is usually sparser than the image, we use a relatively shallow decoder. We demonstrate our method on segmentation of intracranial haemorrhages from computed tomography (CT) head scans and show comparable performance to existing models. On a brain tissue segmentation task on CT images, our method outperformed the existing methods especially on tiny structure.Clinical relevanceCT images are the first line of head scans when stroke or haemorrhage is suspected. Segmentation of intracranial haemorrhages (ICH) can provide clinicians with important measures of brain lesions to decide on treatment procedure or surgical decision. Brain and tissue volumes are indicative of neurodegenerative diseases. The proposed techniques can be used to segment ICH and brain tissues from CT head scans.

show abstract

Section: Related Workmentioning

confidence: 99%

“…We obtained publicly available datasets from INSTANCE'22 challenge [15], [17] and PhysioNet [18]. The former has 893 slices from 100 scans that contain annotated intracranial hemorrhage (ICH) lesions.…”

Section: A1 Ich Lesion Segmentationmentioning

confidence: 99%

Sequential Modelling for Medical Image Segmentation

How,

Rajapakse

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…In terms of my experiments, I performed intracranial hemorrhage lesion segmentation similar to the published INSTANCE challenge [127,128]. All top rank models in the challenge are UNet variant such as ResUNet, nnU-Net and Attention U-Net.…”

Section: Related Workmentioning

confidence: 99%

“….1 Datasets I obtained publicly available annotated ICH datasets from INSTANCE'22 challenge[127,128] and PhysioNet[129]. The former has 893 slices from 100 scans that contain annotated intracranial hemorrhage (ICH) lesions.…”

mentioning

confidence: 99%

Deep learning application on head CT images

How

View full text Add to dashboard Cite

I studied on the deep learning frontier application on head Computed Tomography (CT) scans due to the wide adoption and cost efficiency of CT scans. The works were formulated in 4 chapters; In Chapter 1, I introduced the technical knowledge of CT images and the medical problems being solved using deep learning. In Chapter 2, I dived into the problem of image segmentation for brain intracranial hemorrhage (ICH) with small annotated mask dataset. My proposed training framework Meta Pseudo Segmentation (MPS) trained segmentation model with consistency training and student-teacher learning, outperforming supervised learning and EM algorithm. In Chapter 3, I tackled the problem of pseudo-healthy generation using VQGAN. My method outperformed the previous work substantially in synthesis quality. In Chapter 4, I proposed a unified multi-task segmentation model to perform ICH segmentation and brain tissue segmentation on CT. My model performed best in segmenting complex and granular region.

show abstract

Automated segmentation of ventricular volumes and subarachnoid hemorrhage from computed tomography images: Evaluation of a rule-based pipeline approach

Butler,

Shah,

Ozgen

et al. 2024

Neuroradiol J

View full text Add to dashboard Cite

Changes in ventricular size, related to brain edema and hydrocephalus, as well as the extent of hemorrhage are associated with adverse outcomes in patients with subarachnoid hemorrhage (SAH). Frequently, these are measured manually using consecutive non-contrast computed tomography scans. Here, we developed a rule-based approach which incorporates both intensity and spatial normalization and utilizes user-defined thresholds and anatomical templates to segment both lateral ventricle (LV) and SAH blood volumes automatically from CT images. The algorithmic segmentations were evaluated against two expert neuroradiologists on representative slices from 20 admission scans from aneurysmal SAH patients. Previous methods have been developed to automate this time-consuming task, but they lack user feedback and are hard to implement due to large-scale data and complex design processes. Our results using automatic ventricular segmentation aligned well with expert reviewers with a median Dice coefficient of 0.81, AUC of 0.91, sensitivity of 81%, and precision of 84%. Automatic segmentation of SAH blood was most reliable near the base of the brain with a median Dice coefficient of 0.51, an AUC of 0.75, precision of 68%, and sensitivity of 50%. Ultimately, we developed a rule-based method that is easily adaptable through user feedback, generates spatially normalized segmentations that are comparable regardless of brain morphology or acquisition conditions, and automatically segments LV with good overall reliability and basal SAH blood with good precision. Our approach could benefit longitudinal studies in patients with SAH by streamlining assessment of edema and hydrocephalus progression, as well as blood resorption.

show abstract

The state-of-the-art 3D anisotropic intracranial hemorrhage segmentation on non-contrast head CT: The INSTANCE challenge

Cited by 3 publications

References 38 publications

Sequential Modelling for Medical Image Segmentation

Sequential Modelling for Medical Image Segmentation

Deep learning application on head CT images

Automated segmentation of ventricular volumes and subarachnoid hemorrhage from computed tomography images: Evaluation of a rule-based pipeline approach

Contact Info

Product

Resources

About