Fig. 1. Flow chart of FMT reconstruction algorithm
Fig. 2. Flow diagram of FMT reconstruction algorithm based on volume compensation
Fig. 3. Image reconstruction of double target phantom. (a) Double target phantom; (b) triangular mesh for FEM
Fig. 4. Convergence performance of different clustering algorithm in terms of the cluster uniformity JC
Fig. 5. Convergence performance of different clustering algorithm in terms of the cluster uniformity Jb
Fig. 6. Reconstructed results for double target phantom. Reconstructed results (a) before and (b) after volume compensation
Fig. 7. Reconstruction images before and after the volume compensation when adding Gaussian noises of (a) RSN=40 dB, (b) RSN =30 dB, and (c) RSN =20 dB
Fig. 8. Reconstruction results of triple target phantom. (a) Triple targets phantom; (b) mesh for FEM; reconstructed results (c) before and (d) after volume compensation
Parameter | μaf /mm-1(left/right) | μae /mm-1 | μam /mm-1 | μ'se/μ'sm | γ | η |
---|
Anomaly | 0.04, 0.04 | 0.0025 | 0.003 | 3.0 | -0.431 | 0.16 | Background | 0.0025 | 0.0025 | 0.003 | 3.0 | -0.431 | 0.16 |
|
Table 1. Optical parameters of forward model
Method | MSE /10-4 | Relativeerror /10-2 | CPUtime /s |
---|
Traditionalperturbation | 9.32 | 4.86 | 624 | Proposedalgorithm | 7.61 | 3.82 | 678 |
|
Table 2. Quantitative evaluation of different methods for double target phantom (iterate 60 times)
Method | MSE /10-4 | Relativeerror /10-2 | CPUtime /s |
---|
Traditionalperturbation | 9.57 | 5.02 | 624 | Proposedalgorithm | 7.75 | 3.94 | 678 |
|
Table 3. Quantitative evaluation of different methods for three target phantom (iterate 60 times)