Enter Your Email Address to Receive a
Copy of MedicExhange Member Demograhpics

 

A new application of mpMP4 technology promises to provide an effective faster method of compressing and viewing remotely delivered CT images of injured personnel in combat and possibly other remote environments, according to study data presented at the 2008 meeting of the Radiological Society of North America (RSNA). This study is useful for deployed medical forces in remote locations where CT is used as a highly specific triage tool in critically injured patients. There is difficulty transmitting these large file sizes as the patient is evacuated further due to very limited upstream bandwidth in Iraq and Afghanistan.

“We have to come up with a way to compress the images and send them over low bandwidth connections," reported Paul Gabriel Peterson, MD, Captain, Medical Corps, Diagnostic Radiology Resident, Walter Reed Army Medical Center, Washington, DC who conducted the study with Dr. Binh Nguyen, Sung Pak and Dr. Les Folio. (Read the abstract here.) The question the authors and Dr. Peterson sought to answer was: “Is there a rapid, high fidelity compression method which would enable transmission of medical imaging data over low bandwidth internet connections?”

The study used a small sample of patients (n=25) who had a chest, abdomen, or pelvic CT. Each study was compressed using MP4 technology at 10 frames/second and each study compressed using three separate compression ratios (CRs): 164:1 (250 Kbps), 82:1 (500 Kbps), 41:1 (1000 Kbps). The standard window/level used was W: 1500, L: -200. Compressed videos converted back to DICOM and reloaded onto PACS with three 3 board-certified Radiologists who reviewed each series, from highest to lowest compression, followed by the original scan. The radiologists then evaluated the following for urgent/emergent findings (yes/no): free air, major vascular injury, spinal cord injury/impending injury, rupture of solid organs, free fluid/blood, and foreign bodies in critical locations. In terms of averages, the average file size for the original DICOM image data was 73 MB, the average compressed file sizes were 250Kbps (0.4 MB), 500Kbps (0.8 MB) and 1000Kbps (1.7 MB), and the average CRs were 250 Kbps (171:1), 500 Kbps (86:1), and 1000 Kbps (41:1). Discrepancy between calculated file sizes and actual file sizes was likely due to extraneous data included with DICOM data (non-image data) (770KB vs. 512KB per image). “This compression technique uses ‘literal video rendering ’ and we used only single pass compression to increase speed. Bottom-line, each frame of the video varies in compression, which is why the compression ratios for each study are different; however, the averages do approximate our calculated CRs,” stated Dr. Peterson.

The results had a combined sensitivity (95% CI) and combined specificity (95% CI) for all CRs above 90%. In most cases, CRs approached 100% in terms of specificity. The optimal CRs were 82:1 and 41:1 which gave 94% sensitivity and 100% specificity respectively.  Confidence intervals were calculated using the method of generalized estimating equations to adjust for the lack of independent data (3 reviewers looked at the same 25 scans).

The results were satisfactory for a non-diagnostic image review and suggest that this technology would increase the likelihood priors are available for comparison at tertiary care centers. The optimal CR provided excellent images on both the 500 and 1000 Kbps MP4 compression methods (CRs approximately 82:1 and 41:1, respectively) and resulted in high sensitivity and specificity. “The benefits are that the downrange scans are more likely to follow patients to tertiary care centers, therefore, non-negligible radiation dose becomes more than simply for triage,” Dr. Peterson told MedicExchange.

The limitations in our series were few “fresh” trauma cases with a low number of solid organ injuries, which appear most challenging to see on compressed images, and the limited window/level capability. To address the limitations of few trauma cases, Dr. Peterson said “…this is relatively insignificant since this study was directed towards evaluating the fidelity of compression technique, which should apply to all scans regardless of clinical history.” Also in regards to the low number of solid organ injuries, Dr. Peterson stated “…these injuries are most susceptible to volume averaging with the surrounding organ during compression; this issue requires closer attention in future research.”

In conclusion, "MP4 video compression is a promising technique which could be useful in combat operations where images are acquired in austere locations with limited upstream bandwidth available for medical imaging data transmission," Dr. Peterson told MedicExchange. Overall, it takes on average less than 2 minutes to compress the studies and then transmit them remotely. At the remote locations, the images can be converted back to DICOM and loaded onto PAC workstations for viewing or these video files can be viewed directly on handheld electronic devices and personal computers.  “Radiologists should be aware of this technology, which the military is using to increase the care we provide to injured soldiers as it may have applications in other fields of radiology as well.”

 

 

Download Embed Embed this video on your site