Fig. 1. Principle of the Hessian filter.

Fig. 2. Sensitivity of the Hessian filter to the maximum scale: (a–c) segmentation results obtained by the multiscale Hessian filter using maximum scale 1, 7, 10, and (d–f) is the corresponding vessel diameter quantification map.

Fig. 3. Performance of the multiscale Hessian-based segmentation algorithm: (a) the original photoacoustic microangiography, (b) segmentation result obtained by the local adaptive threshold method, (c) segmentation result obtained by the Hessian filter, (d) segmentation result obtained by the proposed method with an over-large maximum scale, (e) segmentation result obtained by the proposed method with an appropriate maximum scale, and (f) the corresponding vessel diameter quantification map of (e).

Fig. 4. Skeletal image of Fig. 3(a).

Fig. 5. Small area quantification maps: (a) vessel length fraction map, (b) vessel density map, and (c) fractal dimension map.

Fig. 6. Two regions of interest (ROI) selection.

Fig. 7. Mean and standard deviations of the measurement parameters within the two ROIs: (a) vessel diameter, (b) vessel length fraction, (c) vessel density, (d) fractal dimension.