Segmentation of Plantar Foot Thermal Images Using Prior Information

Abstract

Diabetic foot (DF) complications are associated with temperature variations. The occurrence of DF ulceration could be reduced by using a contactless thermal camera. The aim of our study is to provide a decision support tool for the prevention of DF ulcers. Thus, the segmentation of the plantar foot in thermal images is a challenging step for a non-constraining acquisition protocol. This paper presents a new segmentation method for plantar foot thermal images. This method is designed to include five pieces of prior information regarding the aforementioned images. First, a new energy term is added to the snake of Kass et al. in order to force its curvature to match that of the prior shape, which has a known form. Second, we defined the initial contour as the downsized prior-shape contour, which is placed inside the plantar foot surface in a vertical orientation. This choice makes the snake avoid strong false boundaries present outside the plantar region when evolving. As a result, the snake produces a smooth contour that rapidly converges to the true boundaries of the foot. The proposed method is compared to two classical prior-shape snake methods, that of Ahmed et al. and that of Chen et al. A database of 50 plantar foot thermal images was processed. The results show that the proposed method outperforms the previous two methods with a root-mean-square error of 5.12 pixels and a dice similarity coefficient of 94%. The segmentation of the plantar foot regions in the thermal images helped us to assess the point-to-point temperature differences between the two feet in order to detect hyperthermia regions. The presence of such regions is the pre-sign of ulcers in the diabetic foot. Furthermore, our method was applied to hyperthermia detection to illustrate the promising potential of thermography in the case of the diabetic foot. Associated with a friendly acquisition protocol, the proposed segmentation method is the first step for a future mobile smartphone-based plantar foot thermal analysis for diabetic foot patients.

Keywords: active contours; diabetic foot; hyperthermia; infrared camera; plantar foot thermal images; prior-shape-based segmentation.

Figure 4

Segmentation of an IR thermal image with seven blind-segmentation methods: (a) a typical image, (b) region-growing method [25], (c) snake method [24], (d) Chan and Vese method [23], (e) Zhang et al. method [26], (f) Dong et al. method [27], (g) Li et al. method [28] and (h) Ben Salah et al. method [29]. Red curve corresponds to the segmentation results and green one is the ground truth contour.

Figure 7

Segmentation of a plantar foot thermal image with different initial contours. The different positions of the initial contour are in (a–c). Our method’s results are in (d–f). The Ahmed et al. [39] method is (g–i), and the Chen et al. [38] method is (j–l). Blue and green curves correspond respectively to segmentation results and the ground truth contours.

Figure 1

Example of an acquired thermal image.

Figure 2

The FlirOne Pro camera.

Figure 3

Acquired image preprocessing, (a) the two feet, (b) the right foot, (c) the left foot and (d) the flipped left foot.

Figure 5

(a) Ten registered foot contours. (b) The prior-shape contour. (c) The initial contour.

Figure 6

Segmentation of three plantar foot thermal images. The proposed method results are in (a–c). The Ahmed et al. results are in (d–f). The Chen et al. results are in (g–i). Blue and green curves correspond respectively to segmentation results and the ground truth contours.

Figure 8

Thermal difference map $\left|\mathsf{\Delta }T\right|$ for two control subjects.images (a,b) correspond to two healthy persons. Their thermal maps $\left|\mathsf{\Delta }T\right|$ are respectively in (c,d).

Figure 9

Thermal ifference map $\left|\mathsf{\Delta }T\right|$ for two DF subjects with ulcers. images (a,b) correspond to two diabetic subjects with ulcers. Their thermal maps $\left|\mathsf{\Delta }T\right|$ are respectively in (c,d).