10 Layer Segmentation and Analysis for Retina with Diseases
271
Bland-Altman plot analysis were utilized for a performance comparison between the
proposed method and the ground truth.
To assess our experiments, several measures based on the segmented volume
of the EZ disruption including sensitivity (SEN), specificity (SPE) and balanced
accuracy rate (BAR) were adopted. These evaluation indexes are commonly used in
imbalanced classification problems and are defined as below:
S E N
T P
T P + F N
× 100%
(10.11)
S P E
T N
T N + F P
× 100%
(10.12)
B AR
S E N + S P E
2
(10.13)
where TP, FN, TN and FP represent true positive, false negative, true negative and
false negative, respectively.
Figure 10.15 shows one of the detection results using the proposed framework, and
the corresponding ground truth for the EZ disruption region. The en face projections
of the original VOIs, ground truth, and corresponding detected EZ disruption are
also shown. We can see from Fig. 10.15 that while the proposed method detected
the EZ disruption well, there were still some false positives and false negatives. The
detection results for a normal eye are shown in Fig. 10.16. Most of the negative
regions were correctly classified; however, there were still some false positives.
The mean and 95% confidence intervals of the detected disruption volume for the normal eyes were mean normal 0.0037 mm
3 and C I normal
[0.0005, 0.0069] mm
3 , while for the eyes with retinal trauma they were
mean trauma 0.1035 mm
3 and C I trauma [0.0126, 0.1944] mm
3 . The detected
EZ disruption volume comparison between the normal eyes and the eyes with retinal
trauma is shown in Fig. 10.17. Student’s t-test demonstrated a strong statistical significance for the detected EZ disruption volume differences between the two groups
of eyes (p 9.9112 × 10
−8
0.001).
For the eyes with retinal trauma, the SEN was 85.69% ± 9.59%, the SPE was
85.91% ± 5.48%, and the BAR was 85.80% ± 6.16%. For the normal eyes, the SPE
was 99.03% ± 0.73%. Because there were no true positives, the values of SEN and
BAR were irrelevant.
For the eyes with retinal trauma, the correlation between the segmented EZ disruption volume and the ground truth was r = 0.8795 with a significance level p
< 0.0001. The 95% confidence interval for r was 0.6683–0.9595. Figure 10.18 shows
the Bland-Altman plot for the consistency analysis between the automatic segmented
EZ disruption volume and the ground truth.
In summary, in this study, we developed and evaluated an automatic method to
detect the 3D integrity of the EZ in eyes with retinal trauma. Because the disrupted
voxels in the EZ region are much less numerous than the non-disrupted ones, this
leads to a typical imbalanced classification problem. To overcome this problem,
an Adaboost algorithm (at the algorithm level) and dataset balance strategies (at
271
Bland-Altman plot analysis were utilized for a performance comparison between the
proposed method and the ground truth.
To assess our experiments, several measures based on the segmented volume
of the EZ disruption including sensitivity (SEN), specificity (SPE) and balanced
accuracy rate (BAR) were adopted. These evaluation indexes are commonly used in
imbalanced classification problems and are defined as below:
S E N
T P
T P + F N
× 100%
(10.11)
S P E
T N
T N + F P
× 100%
(10.12)
B AR
S E N + S P E
2
(10.13)
where TP, FN, TN and FP represent true positive, false negative, true negative and
false negative, respectively.
Figure 10.15 shows one of the detection results using the proposed framework, and
the corresponding ground truth for the EZ disruption region. The en face projections
of the original VOIs, ground truth, and corresponding detected EZ disruption are
also shown. We can see from Fig. 10.15 that while the proposed method detected
the EZ disruption well, there were still some false positives and false negatives. The
detection results for a normal eye are shown in Fig. 10.16. Most of the negative
regions were correctly classified; however, there were still some false positives.
The mean and 95% confidence intervals of the detected disruption volume for the normal eyes were mean normal 0.0037 mm
3 and C I normal
[0.0005, 0.0069] mm
3 , while for the eyes with retinal trauma they were
mean trauma 0.1035 mm
3 and C I trauma [0.0126, 0.1944] mm
3 . The detected
EZ disruption volume comparison between the normal eyes and the eyes with retinal
trauma is shown in Fig. 10.17. Student’s t-test demonstrated a strong statistical significance for the detected EZ disruption volume differences between the two groups
of eyes (p 9.9112 × 10
−8
0.001).
For the eyes with retinal trauma, the SEN was 85.69% ± 9.59%, the SPE was
85.91% ± 5.48%, and the BAR was 85.80% ± 6.16%. For the normal eyes, the SPE
was 99.03% ± 0.73%. Because there were no true positives, the values of SEN and
BAR were irrelevant.
For the eyes with retinal trauma, the correlation between the segmented EZ disruption volume and the ground truth was r = 0.8795 with a significance level p
< 0.0001. The 95% confidence interval for r was 0.6683–0.9595. Figure 10.18 shows
the Bland-Altman plot for the consistency analysis between the automatic segmented
EZ disruption volume and the ground truth.
In summary, in this study, we developed and evaluated an automatic method to
detect the 3D integrity of the EZ in eyes with retinal trauma. Because the disrupted
voxels in the EZ region are much less numerous than the non-disrupted ones, this
leads to a typical imbalanced classification problem. To overcome this problem,
an Adaboost algorithm (at the algorithm level) and dataset balance strategies (at