Implementation of Deep Learning based Neural Network Algorithm for Intracranial Haemorrhage Detection

Project Title

Implementation of Deep Learning based Neural Network Algorithm for Intracranial Haemorrhage Detection

Project Area of Specialization Artificial IntelligenceProject Summary

Hemorrhage in the head (intracranial hemorrhage) is a relatively common condition that has many causes ranging from trauma, stroke, aneurysm, vascular malformations, high blood pressure, illicit drugs and blood clotting disorders. The neurologic consequences also vary extensively depending upon the size, type of hemorrhage and location ranging from headache to death. The role of the Radiologist is to detect the hemorrhage, characterize the hemorrhage subtype, its size and to determine if the hemorrhage might be jeopardizing critical areas of the brain that might require immediate surgery.
This project introduces the concept of deep learning based automatic intracranial hemorrhage recognition & detection from CT Scans images. Purpose of this is to eliminate role of human from detection purpose to increase efficiency, save time and manpower and to avoid human error. CT Scans used widely in medical for detecting the type of hemorrhage etc and different methods and algorithms are proposed for its detection and classification.
Their accuracy and precision was calculated but not the efficiency, so we will be implementing the algorithm that gives efficiency too along with the accuracy. The task of detecting and identifying hemorrhage is very sensitive as it is a serious health problem requiring often intensive medical treatment. It requires very high precision and accuracy in very short time. Several techniques are available for detection but this thesis focuses on training a deep learning convolutional neural network model that is trained to detect images and identify the type of hemorrhage. CNNs are more useful because they automatically extract features of targets unlike machine learning algorithms.
In this project, we create CNN models for automatic detection based on YOLO. We then try methods to increase metrics like precision, recall and Jaccard Index. We will develop our solution while using the rich image dataset provided by Radiological Society of North America (RSNA) in collaboration with the members of American Association of Neuroradiology and MD.ai. We then annotated it for Darknet YOLO and proceeded towards the classification of the hemorrhage and its subtypes.

 

Project Objectives

The key requirement & objective of this project is to create a model that detects intracranial Hemorrhage and its type from the CT Scans of the brain, with high precision and accuracy.. Hemorrhage detection is not an easy task especially when images are CT scans. A human can perform this task but targets are not clearly identifiable from naked human eye which results in delayed and poor results. Moreover, the identification & classification of Hemorrhage is a slow process. So we will implement the algorithm that will make the process fast.

While all acute (i.e. new) hemorrhages appear dense (i.e. white) on computed tomography (CT), the primary imaging features that help Radiologists determine the subtype of hemorrhage are the location, shape and proximity to other structures. So in this project we will detect and identify the Type and its subtype of Haemorrhage using CT scans using the fast and efficient deep learning algorithm. 

Project Implementation Method

Our task is to create a model for automatic detection of hemorrhage. So, we proposed a deep learning based YOLO model which contains convolutional neural networks. YOLO uses an approach in which we split our images in different regions and then a bounding box is created for each region according to some calculated weights. YOLO is very fast because it processes the whole image at once and some pre-trained models are also available for fine tuning purposes. We are training YOLO model on our dataset provided by Radiological Society of North America (RSNA) . Training results will be shared and in future, we’ll try techniques to improve efficiency of trained models.

Proposed Model

Our goal is to train a classifier which should be fast enough to provide real time hemorrhage detection. We have seen CNNs implemented in previous chapter. They apply filters to an image at multiple locations and high scoring locations are called detections. But we are using a different approach here. We apply a neural network to an image which divides image and different regions and then bounding boxes are predicted for each region. A very similar approach is proposed by a real time detector named You Only Look Once (YOLO). which is our inspiration to perform detection in this way.

YOLO

You Only Look Once (YOLO) is a state of the art, real time object detection system. It applies a single neural network to whole image. This network divides the image in regions and predict bounding boxes for each region. These bounding boxes are weighted by calculated probabilities. It looks at the whole image at training time which makes it extremely fast than other object detection techniques.

Benefits of the Project

Intracranial hemorrhage, bleeding that occurs inside the cranium, is a serious health problem requiring rapid and often intensive medical treatment. Diagnosis requires an urgent procedure. When a patient shows acute neurological symptoms such as severe headache or loss of consciousness, highly trained specialists review medical images of the patient’s cranium to look for the presence, location and type of hemorrhage. The process is complicated and often time consuming. We will implement a new algorithm on CT Scans dataset for detecting the Hemorrhage to increase the efficiency of the output results and we will also make it fast enough to give the output in short time in order to save the life of patient. Development of a deep learning solution is found to be relatively straight-forward, we can view this as a major step forward in the medical computing field.

   (Intracranial Haemorrhage) 

      (Subdural Haemorrhage) 

Technical Details of Final Deliverable

This project will compose of Software System. CT Scan image will be fed to the system. The system will take that image , process it , detect the haemorrhage and then classify it that which haemorrhage. It will involve the training and testing of the dataset of CT Scans. The results can be visualized in the form of written text and picture explaining which type & sub type of haemorrhage exists. The designed architecture of YOLO Algorithm and the designed Layered Structure of Deep Neural Networks acording to the dataset will be designed.  

Final Deliverable of the Project Software SystemCore Industry MedicalOther Industries IT , Health Core Technology Artificial Intelligence(AI)Other Technologies Robotics, NeuroTechSustainable Development Goals Good Health and Well-Being for People, Industry, Innovation and Infrastructure, Partnerships to achieve the GoalRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	76000
Report	Miscellaneous	12	500	6000
HDMI Wire	Equipment	2	2000	4000
Hard Disk (For storing large dataset)	Equipment	3	8000	24000
RAM	Equipment	1	8000	8000
Mouse	Equipment	2	1000	2000
GPU - NIVIDIA Toolkit	Equipment	2	16000	32000