The increasing incidence of diabetes mellitus (DM) in modern society has become a serious issue. DM can also lead to several secondary clinical complications. One of these complications is diabetic retinopathy (DR), which is the leading cause of new cases of blindness for adults in the United States. While DR can be treated if screened and caught early in progression, the only currently effective method to detect symptoms of DR in the eyes of DM patients is through the manual analysis of fundus images. Manual analysis of fundus images is time-consuming for ophthalmologists and can reduce access to DR screening in rural areas. Therefore, effective automatic prescreening tools on a cloud-based platform might be a potential solution to that problem. Recently, deep learning (DL) approaches have been shown to have state-of-the-art performance in image analysis tasks. In this study, we established a research PACS for fundus images to view DICOMized and anonymized fundus images. We prototyped a deep learning engine in the PACS server to perform prescreening classification of uploaded fundus images into DR grade. We fine-tuned a deep convolutional neural network (CNN) model pretrained on the ImageNet dataset by using over 30,000 labeled image samples from the public Kaggle Diabetic Retinopathy Detection fundus image dataset6. We linked the PACS repository with the DL engine and demonstrated the output predicted result of DR into the PACS worklist. The initial prescreened result was promising and such applications could have potential as a “second reader” with future CAD development for nextgeneration PACS.
Retinal changes on a fundus image have been found to be related to a series of diseases. The traditional retinal image quantitative features are usually collected by various standalone and proprietary software which results in variabilities in feature extraction and data collection. Based on our previously established web-based imaging informatics platform to view DICOMized and de-identified fundus images, we developed a computer aided detection structured report (CADe SR) to capture some of the quantitative features on fundus images such as arteriole/venule diameter ratio, cup/disc diameter ratio and to record several lesions such as aneurysms, hemorrhages, neovascularization and exudates into different regions based on known research and clinically related templates such as Early Treatment Diabetic Retinopathy Study (ETDRS) 9 Region Map and four Region Map. In this way, the location patterns of the above lesions as well as morphological changes of anatomy structures could be saved in SR for further radiomics research. In addition, an on-line consultation tool was developed to facilitate further discussion among clinicians and researchers regarding any uncertainty of measurements. Compared with the present workflow of utilizing standalone software to obtain quantitative results, qualitative and quantitative data was acquired by the CADe SR directly, which will provide researchers and clinicians the ability to capture findings and will foster future image-based knowledge discovery researches.
Proc. SPIE. 10138, Medical Imaging 2017: Imaging Informatics for Healthcare, Research, and Applications
KEYWORDS: Medicine, Image compression, Computer aided diagnosis and therapy, Detection and tracking algorithms, Data modeling, Cameras, Databases, Medical imaging, Computer aided design, Algorithm development
Diabetic retinopathy (DR) is one of the serious complications of diabetes that could lead to blindness. Digital fundus camera is often used to detect retinal changes but the diagnosis relies too much on ophthalmologist’s experience. Based on our previously developed algorithms for quantifying retinal vessels and lesions, we developed a computer aided detection-structured report (CAD-SR) template and implemented it into picture archiving and communication system (PACS). Furthermore, we mapped our CAD-SR into HL7 CDA to integrate CAD findings into diabetes patient electronic patient record (ePR) system. Such integration could provide more quantitative features from fundus image into ePR system, which is valuable for further data mining researches.
This paper presents an algorithm to identify the handwritten and the printed texts among document images. The characteristic of stroke thickness is used and a kind of calculating method is designed for this feature. The proposed method, which is clearly defined and easily realized, calculates the stroke thickness feature by counting edge pixels in a neighborhood. Document images are generally divided into text lines or characters. However, the line and the character are not conducive to the judgment between handwritten and printed text distinction. The line is too rough and the character is too small. Using the stroke thickness characteristics, combined with layout analysis, the text line in the document image is further divided into the area of uniform thickness. This kind of area is more detailed than text line and larger than a single character. So more stable features can be extracted from it. Last, the features of these regions are divided by using SVM. The proposed algorithm obtained better performance in the document image database including handwritten and printed texts.
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.
This paper proposed a convenient navigation system for neurosurgeon's 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 neurosurgeon's pre-operative planning and teaching.