Xingyun Geng

Computer aided quantification for retinal lesions in patients with moderate and severe non-proliferative diabetic retinopathy: a retrospective cohort study

BackgroundDetection of retinal lesions like micro-aneurysms and exudates are important for the clinical diagnosis of diabetes retinopathy. The traditional subjective judgments by clinicians are dependent on their experience and can be subject to lack of consistency and therefore a quantification method is worthwhile.MethodsIn this study, 10 moderate non-proliferative diabetes retinopathy (NPDR) patients and 10 severe NPDR ones were retrospectively selected as a cohort. Mathematical morphological methods were used for automatic segmentation of lesions. For exudates detection, images were pre-processed with adaptive histogram equalization to enhance contrast, then binary images for area calculation were obtained by threshold classification. For micro-aneurysms detection, the images were pre-processed by top-hat and bottom-hat transformation, then Otsu method and Hough transform were used to classify micro-aneurysms. Post-processing morphological methods were used to preclude the false positive noise.ResultsAfter segmentation, the area of exuduates divided by optic disk area (exudates/disk ratio) and counts of microaneurysms were quantified and compared between the moderate and severe non-proliferative diabetic retinopathy groups, which had significant difference(P < 0.05).ConclusionsIn conclusion, morphological features of lesion might be an image marker for NPDR grading and computer aided quantification of retinal lesion could be a practical way for clinicians to better investigates diabetic retinopathy.

Interoperative fundus image and report sharing in compliance with integrating the healthcare enterprise conformance and web access to digital imaging and communication in medicine persistent object protocol

AIM To address issues in interoperability between different fundus image systems, we proposed a web eye-picture archiving and communication system (PACS) framework in conformance with digital imaging and communication in medicine (DICOM) and health level 7 (HL7) protocol to realize fundus images and reports sharing and communication through internet. METHODS Firstly, a telemedicine-based eye care work flow was established based on integrating the healthcare enterprise (IHE) Eye Care technical framework. Then, a browser/server architecture eye-PACS system was established in conformance with the web access to DICOM persistent object (WADO) protocol, which contains three tiers. RESULTS In any client system installed with web browser, clinicians could log in the eye-PACS to observe fundus images and reports. Multipurpose internet mail extensions (MIME) type of a structured report is saved as pdf/html with reference link to relevant fundus image using the WADO syntax could provide enough information for clinicians. Some functions provided by open-source Oviyam could be used to query, zoom, move, measure, view DICOM fundus images. CONCLUSION Such web eye-PACS in compliance to WADO protocol could be used to store and communicate fundus images and reports, therefore is of great significance for teleophthalmology.

A self-adaptive distance regularized level set evolution method for optical disk segmentation

The optic disc (OD) is one of the important anatomic structures on the retina, the changes of which shape and area may indicate disease processes, thus needs computerized quantification assistance. In this study, we proposed a self-adaptive distance regularized level set evolution method for OD segmentation without the periodically re-initializing steps in the level set function execution to a signed distance function during the evolution. In that framework, preprocessing of an image was performed using Fourier correlation coefficient filtering to obtain initial boundary as the beginning contour, then, an accurate boundary of the optic disc was obtained using the self-adaptive distance regularized level set evolution method. One hundred eye fundus color numerical images from public database were selected to validate our algorithm. Therefore, we believe that such automatic OD segmentation method could assist the ophthalmologist to segment OD more efficiently, which is of significance for future computer-aided early detection of glaucoma and retinopathy diseases.

The Development of a Smart Personalized Evidence Based Medicine Diabetes Risk Factor Calculator

Type 2 diabetes mellitus (T2DM) is a chronic disease affected with complex risk factors and has been regarded as one of the major social burdens due to its high occurrence. In this study, we aim to incorporate the idea of evidence based medicine (EBM) into our diabetes risk factor App development. We acquired and extracted the relative risk of different risk factors from relevant literature by searching academic databases. A total of 19 items of risk factors in our daily lives has been finally selected. To design App graphic interface, a total of three pages were designed to let user answer the questions and show the results of their level or risk to have T2DM. We validated the feasibility of our App in 100 users and the results were promising. Therefore, the personalized EBM diabetes risk factor calculator might be a feasible approach to remind those T2DM risky populations by revealing their potential risk factors, thus making implementation of personalized and prevention medicine achievable at hand.

Clinic expert information extraction based on domain model and block importance model

To extract expert clinic information from the Deep Web, there are two challenges to face. The first one is to make a judgment on forms. A novel method based on a domain model, which is a tree structure constructed by the attributes of query interfaces is proposed. With this model, query interfaces can be classified to a domain and filled in with domain keywords. Another challenge is to extract information from response Web pages indexed by query interfaces. To filter the noisy information on a Web page, a block importance model is proposed, both content and spatial features are taken into account in this model. The experimental results indicate that the domain model yields a precision 4.89% higher than that of the rule-based method, whereas the block importance model yields an F1 measure 10.5% higher than that of the XPath method.

Optical augmented reality assisted navigation system for neurosurgery teaching and planning

This paper proposed a convenient navigation system for neurosurgeons pre-operative planning and teaching with augmented reality (AR) technique, which maps the three-dimensional reconstructed virtual anatomy structures onto a skull model. This system included two parts, a virtual reality system and a skull model scence. In our experiment, a 73 year old right-handed man initially diagnosed with astrocytoma was selected as an example to vertify our system. His imaging data from different modalities were registered and the skull soft tissue, brain and inside vessels as well as tumor were reconstructed. Then the reconstructed models were overlayed on the real scence. Our findings showed that the reconstructed tissues were augmented into the real scence and the registration results were in good alignment. The reconstructed brain tissue was well distributed in the skull cavity. The probe was used by a neurosurgeon to explore the surgical pathway which could be directly posed into the tumor while not injuring important vessels. In this way, the learning cost for students and patients’ education about surgical risks reduced. Therefore, this system could be a selective protocol for image guided surgery(IGS), and is promising for neurosurgeons pre-operative planning and teaching.

Biological fiducial point based registration for multiple brain tissues reconstructed from different imaging modalities

With the development of computer aided navigation system, more and more tissues shall be reconstructed to provide more useful information for surgical pathway planning. In this study, we aimed to propose a registration framework for different reconstructed tissues from multi-modalities based on some fiducial points on lateral ventricles. A male patient with brain lesion was admitted and his brain scans were performed by different modalities. Then, the different brain tissues were segmented in different modality with relevant suitable algorithms. Marching cubes were calculated for three dimensional reconstructions, and then the rendered tissues were imported to a common coordinate system for registration. Four pairs of fiducial markers were selected to calculate the rotation and translation matrix using least-square measure method. The registration results were satisfied in a glioblastoma surgery planning as it provides the spatial relationship between tumors and surrounding fibers as well as vessels. Hence, our framework is of potential value for clinicians to plan surgery.

Chniese document classification using field association knowledge base

Field Association (FA) terms are a limited set of discriminating terms that offer human knowledge to identify document (text) fields. Field association knowledge base (FAKB) is composed of FA terms and their potential hierarchical relationship of the fields belongs to. The primary goal of this research is to build a system that can imitate the process whereby humans recognize the fields by looking at a few Chinese FA terms in a document (text). The documents classification experiment is made on two data collections under different circumstances, including 4000 and 1300 documents respectively. FAKB outperforms all the other statistical methods (SVMs, kNN, and NB) with the average accuracies of 97.7% and 89%. All the experimental results clearly prove that the presented novel method is effective in Chinese document classification.

Registration and fusion of human craniocerebral 3d models reconstructed from multimodality images

Our aim was to investigate the registration and fusion of models reconstructed from multimodal craniocerebral sectional images, providing basis of craniocerebral structure visualization for surgical navigation. The craniocerebral dataset from different modalities were obtained and constructed. For segmentation, “training area” and “threshold” methods were performed for white matter in MRI coronary sections, gray matter and ventricle in MRI axial sections in 3D-DOCTOR4.0, then manual corrected. Skull in CT axial sectional images was automatically segmented by thresholding in Mimics8.11. Then, the above segmented structures were reconstructed and exported to Mimics software for registration and fusion. Finally, the fused models were observed at random angle and measured in computer. Skull, gray matter, white matter and ventricular system were reconstructed. The fused brain models were distributed well in cranial cavity, and inner white matter and ventricles were fused well in good position. In conclusion, three dimensional reconstructions can be made from multimodal craniocerebral sectional images according to their anatomic structure characteristics, and then registered and fused together.