Effects of a digital filter on detectability of a phantom lesion in a scintigram of the equine tarsus.

The research explores how a digital filter impacts the detection of artificial lesions in the scintigrams of horse’s ankle joint. It concludes that the use of a filter can enhance sensitivity in certain images, although it may also increase false positives.

Research Objective and Methodology

• The investigators aimed to investigate the effect of a digital filter on the detectability of artificial (or phantom) lesions in a scintigram, a type of medical imaging, of the equine tarsus, which is the horse’s ankle joint.
• The researchers created a phantom lesion of varying intensity (0, 10, 15, 20, or 30% increased intensity compared to the local background) and placed it in the centrodistal tarsal joint area in the images.
• A total of five different images, each having varying levels of kcounts (kcounts refer to kilocounts per pixel, a measure of the concentration of the radioactive tracer), were obtained. These image sets were at 50, 100, 150, 500, or 1000 kcounts.
• These images were then processed through a Metz filter to observe the effect. A secondary set of the images was kept as is or unfiltered.
• In total, the researchers had 180 images which were projected to a group of nine observers for evaluation.

Findings

• The research established that the application of the digital filter improved the sensitivity, or the true positive rate, in images with lower count levels of 50 and 100 kcounts.
• However, the investigators also noticed an increase in the false positive fraction, or the rate at which normal areas were incorrectly identified as having lesions, at all count levels. These false positives were considerably higher in images with 500 and 1000 kcounts, the two highest levels used in this study.
• From a diagnostic standpoint, an increase in false positives could lead to unnecessary investigations or treatments and cause anxiety in patients. This is an essential factor to consider when applying digital filters to scintigrams for clinical assessments.
• Yet, based on the receiver operating characteristic analysis, a statistical method of measuring the accuracy of a diagnostic test, the filtering of a 50 kcount image improved image quality to the level of a non-filtered 150 kcount image. This finding suggests that digital filtering might contribute to reducing radiation exposure by creating high-quality images from lower kcount levels.