Alamgir Hossain

An intelligent mobile-based automatic diagnostic system to identify retinal diseases using mathematical morphological operations

Diabetic retinopathy is considered in terms of the presence of exudates which cause vision loss in the areas affected. This study targets the development of an intelligent mobile-based automatic diagnosis integrated with a microscopic lens to identify retinal diseases at initial stage at any time or place. Exudate detection is a significant step in order obtaining an early diagnosis of diabetic retinopathy, and if they are segmented accurately, laser treatment can be applied effectively. Consequently, precise segmentation is the fundamental step in exudate extraction. This paper proposes a technique for exudate segmentation in colour retinal images using morphological operations. In this method, after pre-processing, the optic disc and blood vessels are isolated from the retinal image. Exudates are then segmented by a combination of morphological operations such as the modified regionprops function and a reconstruction technique. The proposed technique is verified against the DIARETDB1 database and achieves 85.39% sensitivity. The proposed technique achieves better exudate detection results in terms of sensitivity than other recent methods reported in the literature. In future work, our system will be deployed to a mobile platform to allow efficient and instant diagnosis.

A Cellular Automaton Model of the Effects of Maspin on Cell Migration

Maspin (Mammary Serine Protease Inhibitor) is a non-inhibitory serpin with multiple cellular effects that is a type II tumour metastasis suppressor. Maspin has been shown to reduce cell migration, invasion, proliferation and angiogenesis, and increase apoptosis and adhesion. In this paper, we report the development of a mathematical model of the effects of maspin on cellular proliferation and migration. An artificial neural network has been used to model the unknown cell signalling to determine the cells fate. Results show that maspin reduces migration by between 10-35%; confirmed by published in vitro data. From our knowledge, this is the first attempt to model maspin effects in a computational model to verify in vitro data. This will provide new insights into to the tumour suppressive properties of maspin and inform the development of novel cancer therapy.

Intelligent Appearance and shape based facial emotion recognition for a humanoid robot

In this paper, we present an intelligent facial emotion recognition system with real-time face tracking for a humanoid robot. The system is able to detect facial actions and emotions from images with up to 60 degrees of pose variations. We employ the Active Appearance Model to perform real-time face tracking and extract both texture and geometric representations of images. A POSIT algorithm is also used to identify head rotations. The extracted texture and shape features are employed to detect 18 facial actions and seven basic emotions. The overall system is integrated with a humanoid robot platform to further extend its vision APIs. The system is proved to be able to deal with challenging facial emotion recognition tasks with various pose variations.

An intelligent decision support system for personalized cancer treatment

Cancer is one of the biggest killers in the western world; every two minutes someone is diagnosed with cancer in the UK. Personalized treatment of cancer, which simply means selecting a treatment best suited to an individual involving the integration and translation of several new technologies in clinical care of patients. Conventional cancer treatments include surgery, radiotherapy and chemotherapy. Among these, therapeutically treatment requires optimal control of radiation/drug to minimize toxic effect and in turn to minimize side effect. We propose a hybrid prediction model consist of avascular tumour growth model from a tumour image and intelligent drug scheduling schema for drug penetration. Our main aim is to develop an intelligent decision support system which helps to analyze the tumour microenvironment constraints like cell-cell adhesion, cell movement, extra-cellular matrix (ECM) and optimal solutions of drug scheduling problem. Hypoxia and drug resistance are also incorporated in the model to achieve the predictive results for every patient as both of them considered as the main reason for chemotherapy and radiotherapy treatment failure. Finally, our goal is to provide a dynamic and effective personalized cancer treatment model to support the oncologist for making right decisions to the right patient at the right time.

Red Blood based disease screening using marker controlled watershed segmentation and post-processing

Cell segmentation is a challenging problem due to the complexity and nature of the blood cells. Traditional methods of counting the cells are slow, error prone and often influenced by the performance of the operator. This paper aims to segment and count Red Blood Cells (RBCs) automatically shown in microscopic blood images to determine the condition of the person under examination. We also aim to increase the accuracy of segmentation by precisely looking into the counting of the overlapped cells which is the most conventional challenging task faced by many researchers. The RBCs in this paper are segmented using the integration of marker controlled watershed segmentation with morphological operations. The result of the proposed algorithm was validated with the manual counting method, and a good conformity of about 93.13 % was obtained. The future work will involve segmentation of more complex overlapping cells and the development of Smartphone based realtime disease screening system.

Fractal and image analysis of cytoskeletal changes in tumour cells due to the effects of maspin

Maspin (SERPINB5) is a type II metastasis suppressor that influences multiple cellular functions. To date, maspin has been shown to increase adhesion and apoptosis and to decrease cell migration, proliferation, invasion and metastases in tumour malignancy. At the subcellular level, maspin influences morphological changes in the cell cytoskeleton which regulates complex biological processes including cell migration, cell adhesion and EMT (epithelial to mesenchymal transition). Here non-Euclidian fractal and image analyses have been applied to measure changes in the actin cytoskeleton using confocal microscopy images to confirm the effects of maspin. Results show that maspin contributes to maintaining the regular epithelial like shape, increases cell-cell adhesion and restricts tumour cells from showing the pre-migration and EMT characteristics. Characterization of these changes in the actin cytoskeleton using microscopic image analysis will establish maspin as a potential prognostic marker in future.

A cellular automaton model for hypoxia effects on tumour growth dynamics

Cancer is one of the biggest killers in the western world; every two minutes someone is diagnosed with cancer in the UK. Tumour growth and progression is a complex biological process, normally beginning with genetic mutations in a single cell. It starts with the early or avascular phase where growth is limited by nutrient diffusion, then the vascular stage where angiogenesis occurs to stimulate blood vessel production by the secretion of tumour angiogenesis factors and finally the metastasitic phase where the tumour spreads from the site of origin to distant sites around the body. While considering these events at the cellular level, these processes involve many microenvironment parameters like oxygen concentration, hypoglycaemia, acidity, hypoxia (lack of oxygen), cell-cell adhesion, cell migration and cell-extracellular matrix interactions. In this paper, a computational model is proposed which considered hypoxia as a microenvironment constraint of tumour growth. The model is built on two dimensional cellular automata grid and artificial neural network is considered for establishing signaling network of tumour cells. Each tumour cell can take its own decision in this model. A hypoxia impact was implemented in the model by varying different oxygen concentrations. The results show that hypoxia was introduced in the tumour mass due to lack of oxygen. The model measured tumour invasion and the number of apoptotic cells to support that hypoxia has a critical impacts on avascular tumour growth. This model could inform a better understanding of the impacts of hypoxia in tumour growth from the computational point of view.

Periodic Chemotherapy Dose Schedule Optimization Using Genetic Algorithm

This paper presents a design method for optimal cancer chemotherapy schedules using genetic algorithm (GA). The main objective of chemotherapy is to reduce the number of cancer cells or eradicate completely, if possible, after a predefined time with minimum toxic side effects which is difficult to achieve using conventional clinical methods due to narrow therapeutic indices of chemotherapy drugs. Three drug scheduling schemes are proposed where GA is used to optimize the doses and schedules by satisfying several treatment constraints. Finally, a clinically relevant dose scheme with periodic nature is proposed. Here Martin’s model is used to test the designed treatment schedules and observe cell population, drug concentration and toxicity during the treatment. The number of cancer cells is found zero at the end of the treatment for all three cases with acceptable toxicity. So the proposed design method clearly shows effectiveness in planning chemotherapy schedules.

Neuro-PID adaptive control scheme for blood pressure regulation

Control of physiological states such as mean arterial pressure (MAP) has been successfully achieved using single drug by different control algorithms. Multi-drug delivery demonstrates a significantly challenging task as compared to control with a single-drug. Also the patients sensitivity to the drugs varies from patient to patient. Therefore, the implementation of adaptive controller is very essential to improve the patient care in order to reduce the workload of healthcare staff and costs. This paper presents the design and implementation of a Proportional Integral Derivative controller (PID) using neural network based parameter tuning mechanism to regulate mean arterial pressure and cardiac output (CO) by administering vasoactive and inotropic drugs that are sodium nitroprusside (SNP) and dopamine (DPM) respectively. The parameters of PID controller were optimised offline using Simulink Response Optimization tool. The proposed Neuro-PID controller has been implemented, tested and verified to demonstrate its merits and capabilities as compared to the existing approaches to cover wide range of patients.