Intelligent imaging informatics technology developer Guardian Technologies International is looking for ways to apply its Signature Mapping technology to medical imaging applications. The company is running a series of broad pilot studies involving multiple modalities and multiple diseases through a collaborative effort with the Image Processing and Informatics Laboratory (IPI) at the University of Southern California. IPI provided clinical cases including radiographs with confirmed diagnoses, as well as medical imaging informatics expertise.
For the project, the initial pilot study chosen was multiple sclerosis (MS). To help treat the disease, MS patients undergo multiple MRI scans that require the radiologist to quantify and report any changes to MS lesions over multiple studies and time. This evaluation process is a time-consuming and imprecise activity which could benefits from the automated detection technology such as Signature Mapping, Guardian said.
The MS study found that Signature Mapping algorithms are capable of accurately detecting lesions and, more importantly, provide accurate measurements of size and overall lesion volumes. Compared to clinical observers, Signature Mapping proved to be a more sensitive tool for detecting lesions that were considered marginal or undetectable and provided extremely accurate measurements, while reducing the radiologist analysis time to just seconds.
Normal pressure hydrocephalus was also analyzed by IPI. In the normal course of aging the adult brain begins to shrink as a result the internal ventricles which contain cerebral spinal fluid begin to enlarge. Enlargement of the ventricles is also associated with numerous anomalies including obstructing tumors. The challenge faced by the radiologist is to determine whether the pressure changes in the ventricles are caused by the normal course of aging or as a result of an anomaly. Signature Mapping technologies detected and quantified the ventricles and cranial spinal fluid visualized from the MRI images. IPI reviewed the results of utilizing Signature Mapping and determined the effects to be impressive by demonstrating its ability to detect cerebral spinal fluid, provide a methodology for segmenting the ventricles and determining and quantifying ventricular size and volume.
Acute intracranial hemorrhage was the third area of application in the brain for Signature Mapping. The goal was to develop a Signature Mapping technology for accurately segmenting and detecting intracranial bleeding using axial CT slices. Cases were categorized into three groups; gross identifiable bleeding, subtle small subdural bleeds and extremely small subdural bleeds which could be confused with CT bone hardening artifacts. Signature Mapping demonstrated in all cases to be accurate for the detection of acute intracranial bleeds. It also demonstrated an ability to differentiate between small intracranial bleeds such as less than 5 percent subdural hematomas and difficult to discern bone hardening artifacts typical in most CT scans.
The use of Signature Mapping to monitor tuberculosis (TB) changes over treatment time using radiographic chest x-rays was also studied. A Signature Mapping technology was developed to detect lung volume area, quantifying normal lung volume from diseased lung volume and quantitatively report changes in lung volume through drug treatment progress. Signature Mapping was successful in detecting lung area, quantifying normal lung volumes, and reporting lung volume changes over time during drug treatment therapies. Additionally, it demonstrated itself to be a strong aid in quantification and tracking of TB therapy.
The last and most difficult area that was evaluated was breast imaging. Two specific challenges are in the areas of image clarification and the clinical interpretation of dense breast digital x-ray examinations. Clinicians at USC indicated that as high as 50 percent of their dense breast tumors could be palpated, but were not clearly seen or detected on x-rays. Typically these dense breast areas obscure the visualization of anomalies and potential tumors. Signature Mapping technology was used to clarify difficult-to-visualize areas of the breast and to provide improved visualization and detection of tumor areas and edges. Preliminary results suggest a likely opportunity for early detection and mapping of tumors and tumor growth over time. Microcalcifications were easily visualized and demonstrated unique signature