Use of image transformation to track the natural history of diseases.

The inability to precisely measure the area of a lesion can impair the study of its natural history and response to therapy. This is especially true in case of gastrointestinal lesions, where a standard imaging technique is endoscopy and an unaided visual interpretation of the size of the lesion is difficult. This study presents a novel technique that provides precise measurement of 2-dimensional lesions in cylinder-like organs.

The inability to precisely measure the area of a lesion can impair the study of its natural history and response to therapy. This is especially true in case of gastrointestinal lesions, where a standard imaging technique is endoscopy. In endoscopy, the perceived size of a structure is mostly a function of the distance to the lens. The purpose of this study is to discuss a novel technique (US patent number 7,011,625) that translates and stacks sequential endoscopic images into 2-dimensional maps, where areas can be precisely measured and compared over time. Although the process has been used in 3 clinical trials (2 in the United States and 1 in China), it is not yet commercially available.

The technique is best described graphically. If one thinks of the structure as essentially tubular, then conceptually, the process is to slice and then “flatten” a section of the tube. A series of photographs is taken as the scope moves up or down, and the resulting rectangular regions are then tied together as is depicted in Figure 1.

Figure 1

Image transformation and stacking.

The mathematics necessary to convert the images is essentially a transformation1 (adjusted both for lens characteristics and for mapping of angle measurements to distances) from polar to rectangular coordinates:

It has been shown that the technique provides accurate area measurements and is replicable.2,3

In the Chemoprevention for Barrett’s Esophagus Trial4 (a multisite clinical trial to test whether a COX-2 inhibitor would prevent esophageal cancers in patients with Barrett’s esophagus), the technique was used to analyze changes in lesion size over time.4,5 Figure 2 shows the images of the transformations. Two series of 3 images are shown. The first row was taken at baseline and the second row was taken at 48-week follow-up. The baseline and 48-week follow-up images were of the same region in the esophagus. There are 3 columns. The first is the raw photo, the second is the photo with the structures to be measured outlined, and the third is the images transformed into 2-dimensional maps allowing for quantitative measurement of the surface area of Barrett’s esophagus. The dark areas are the result of the lens of the camera’s lack of alignment to the center of the lumen and are mappings of regions outside of the boundaries of the original picture. Although it was not possible to demonstrate a chemopreventive drug effect using the standard techniques, an alternative and novel analysis6 did show significance.

Figure 2

Barrett’s esophagus transformations.

In an unpublished randomized control ablative therapy trial in patients with Barrett’s esophagus, the techniques were again used (Figure 3). In this case, although ablation significantly reduced the lesion, there was no effect using a chemopreventive agent.

Figure 3

Ablative treatment.

The previous 2 examples are ones where the goal was to accurately measure the change in the size of a precancerous lesion over time. The same process can be used to measure the shape and size of a tissue that has been excised. Figure 4 shows the reconstruction of an excised squamous cell esophageal carcinoma. This analysis is from an unpublished study conducted in China.

Figure 4

Excised lesion.

Although the above examples are all involved with the esophagus, the technique applies to any hollow cylindrical structure into which one can place and align a camera. Figure 5 shows the images of the colon from a routine colonoscopy that have been transformed.

Figure 5

Image transformation in the colon.