Image Content Extraction: Application to MR Images of the Brain

Abstract

A system for automatically extracting image content features was developed that combines registration to a labeled atlas with natural language processing of free-text radiology reports. The system was then tested with T1-weighted, spoiled gradient-echo magnetic resonance (MR) imaging studies of the brain performed in nine patients. The locations of 599 structures were visually assessed by an experienced radiologist and compared with the locations indicated by automated output. The in-plane accuracy of the contours was subjectively evaluated as either good, moderate, or poor. The criterion for classifying a structure as correctly located was that 90% or more of all the images containing the structure had to be correctly identified. For 98% of the structures, the images identified by the automated algorithm agreed with those identified by the radiologist, and in 83% of cases, image contours showed a good in-plane overlap. The results of this validation study demonstrate that this combination of registration and natural language processing is accurate in identifying relevant images from brain MR imaging studies. However, the range of applicability of this technique has yet to be determined by applying the technique to a large number of studies.

A Review of Medical Imaging Informatics

“…We have developed a novel method for image study summarization with specific application to brain MR studies 17,18. Our approach combines expert information from free text radiology reports with an automated registration algorithm that aligns the current patient images to a labeled brain atlas.…”

Natural Language Processing for Biosurveillance

“…Over the last few decades researchers have actively applied NLP techniques to the medical domain (Friedman andHripcsak, 1999, Spyns, 1996). NLP techniques have been used for a variety of applications, including quality assessment in radiology (Fiszman et al, 1998, Chapman et al, 2001b; identification of structures in radiology images (Sinha et al, 2001a, Sinha et al, 2001b; facilitation of structured reporting (Morioka et al, 2002, Sinha et al, 2000 and order entry , Lovis et al, 2001; encoding variables required by automated decision-support systems such as guidelines (Fiszman and Haug, 2000), diagnostic systems (Aronsky et al, 2001), and antibiotic therapy alarms ; detecting patients with suspected tuberculosis (Jain et al, 1996, Knirsch et al, 1998; identifying findings suspicious for breast cancer (Jain and Friedman, 1997), stroke (Elkins et al, 2000), and community acquired pneumonia (Friedman et al, 1999b); and deriving comorbidities from text (Chuang et al, 2002).…”

Correlation between Quantitative and Qualitative Analysis on Image Quality of Digital Dental X-ray Images

“…Due to that, the number of them usually is small. Usually, the surveys included one radiologist [19], three to five dentists [2][3][10] [14] and postgraduates with experience in oral and maxillofacial radiology including digital radiography [4]. The number of image samples used ranges from 12 to 42 panoramic radiographs [1], periapical digital radiographs [2] [4], interproximal radiographs [3] and bitewing [4] images.…”

“…However, preliminary results by Sund and Moystad suggested that the five-point scale was not warranted [10]. Due to that, a three-point scale is common and has been utilized in many surveys related to medical imaging [19][20][21][22][23]. Usually the radiologists were asked to assess the quality of the images before and after certain techniques had been applied [8][9][10][11][12].…”