Automated vertebral body image segmentation for medical screening
A method is disclosed for fully automated segmentation of human vertebral body images in a CT (computerized tomography) study with no user interaction and no phantoms, which has resiliency to anatomical abnormalities, and protocol and scanner variations. The method was developed to enable automated detection of osteoporosis in CT studies performed for other clinical reasons. Testing with 1,044 abdominal CTs from multiple sites, resulted in detection of 96.3% of the vertebral bodies and 1% false positives. Of the detected vertebral bodies, 83.3% were segmented adequately for sagittal plane quantitative evaluation of vertebral fractures indicative of osteoporosis. Improved results were observed when selecting the best sagittal plane of 3 for each vertebra, yielding a segmentation success rate of 85.4%. The method is preferably implemented in software as a building block in a system for automated osteoporosis detection.
1. A method performed in a computer system comprising the steps of:
receiving as an input a digital representation of an image of a volume region of a patient's body containing the spine;
using at least one of statistical and heuristic computational techniques on the digital representation, isolating the spine from the surrounding volume, and in the spine, segmenting the vertebral body trabecular bones, the vertebral body cortexes, and the intervertebral discs, and locating the planes of the intervertebral discs;
wherein segmenting the intervertebral discs includes steps (i) and (ii):
(i) positioning a rectangular parametric curve along the expected center of a vertebral body; and
(ii) inspecting a cross section of the parametric curve; and
creating a graph of weight as a function of position along the spine; and
generating a map corresponding to the volume containing plural elements wherein each element is identified as background or as part of a specified part of the spine.
2. The method of claim 1 wherein the step of locating the planes of the intervertebral discs comprises the step of, for each point in the graph, calculating a moving average, moving maximum, and moving minimum for a moving window typical of the height of a vertebra and its adjacent discs.
3. The method of claim 2 further comprising assigning each weight the first appropriate classification from the following list:
Extreme Spike: weights over a multiple of the moving maximum;
Spike: weights over a multiple of the moving maximum;
High: weights over a multiple of the moving average;
Min: weights at the moving minimum with a small threshold;
Depression: weights below a threshold above the moving minimum;
Above average: weights over the moving average;
Below average: weights that do not fall into any of the classifications above.
4. The method of claim 3 , further comprising evaluating the weights with each of the rules below in the order listed:
Rule ES-M: a minimum between 2 close spikes or extreme spikes identifies a disc, however, if two discs are identified close to each other, the wider of the two does not identify a disc;
Rule ES-MD: in the vicinity of a missing disc, a narrow depression or min between 2 close spikes or extreme spikes identifies a disc;
Rule ESH-MD: in the vicinity of a missing disc, a narrow depression or min between 2 close spikes or extreme spikes or highs identifies a disc;
Rule ESHA-M: in the vicinity of a missing disc, a narrow min between 2 close spikes or extreme spikes, highs, or above averages identifies a disc;
Rule ESHA-MD: in the vicinity of a missing disc, a narrow min or depression between 2 close spikes or extreme spikes, highs, or above averages identifies a disc;
Rule ES-MDB: in the vicinity of a missing disc, a narrow below average, depression or min between 2 close spikes or extreme spikes identifies a disc;
Rule ES-MDBA: in the vicinity of a missing disc, a narrow min, depression, below or above average between 2 close spikes or extreme spikes identifies a disc;
Rule ESH-MDB: in the vicinity of a missing disc, a narrow min, depression, or below average between 2 close spikes or extreme spikes or highs identifies a disc;
Rule ESH-MDBA: in the vicinity of a missing disc, a narrow min, depression, below or above average between 2 close spikes or extreme spikes or highs identifies a disc;
Rule E-SHMDBA: in the vicinity of a missing disc, a narrow min, depression, below, above average, highs, or spikes between 2 extreme spikes identifies a disc;
Rule ESH-MDBA-End: in the vicinity of a missing end disc, a narrow min, depression, below or above average at the end by a close spike or high identifies a disc.
5. The method of claim 4 wherein, after each rule is evaluated, a gap analysis is performed to locate suspect missing discs, identified by gaps between discs that exceed a threshold.
6. The method of claim 4 wherein the rule set is executed repeatedly, until no new discs are detected.
7. The method of claim 1 further comprising calculating a weight using an additive combination of medium density and high density components.