1. |
- Andersén, Christoffer, 1991-, et al.
(author)
-
Deep learning based digitisation of prostate brachytherapy needles in ultrasound images
- 2020
-
In: Medical physics. - : Wiley-Blackwell Publishing Inc.. - 2473-4209 .- 0094-2405. ; 47:12, s. 6414-6420
-
Journal article (peer-reviewed)abstract
- PURPOSE: To develop, and evaluate the performance of, a deep learning based 3D convolutional neural network (CNN) artificial intelligence (AI) algorithm aimed at finding needles in ultrasound images used in prostate brachytherapy.METHODS: Transrectal ultrasound (TRUS) image volumes from 1102 treatments were used to create a clinical ground truth (CGT) including 24422 individual needles that had been manually digitised by medical physicists during brachytherapy procedures. A 3D CNN U-net with 128x128x128 TRUS image volumes as input was trained using 17215 needle examples. Predictions of voxels constituting a needle were combined to yield a 3D linear function describing the localisation of each needle in a TRUS volume. Manual and AI digitisations were compared in terms of the root-mean-square distance (RMSD) along each needle, expressed as median and interquartile range (IQR). The method was evaluated on a dataset including 7207 needle examples. A subgroup of the evaluation data set (n=188) was created, where the needles were digitised once more by a medical physicist (G1) trained in brachytherapy. The digitisation procedure was timed.RESULTS: The RMSD between the AI and CGT was 0.55 (IQR: 0.35-0.86) mm. In the smaller subset, the RMSD between AI and CGT was similar (0.52 [IQR: 0.33-0.79] mm) but significantly smaller (p<0.001) than the difference of 0.75 (IQR: 0.49-1.20) mm between AI and G1. The difference between CGT and G1 was 0.80 (IQR: 0.48-1.18) mm, implying that the AI performed as well as the CGT in relation to G1. The mean time needed for human digitisation was 10 min 11 sec, while the time needed for the AI was negligible.CONCLUSIONS: A 3D CNN can be trained to identify needles in TRUS images. The performance of the network was similar to that of a medical physicist trained in brachytherapy. Incorporating a CNN for needle identification can shorten brachytherapy treatment procedures substantially.
|
|
2. |
- Andersén, Christoffer, 1991-, et al.
(author)
-
Presults for the aI-Brachy study : Utilizing deep learning for needle reconstruction in prostate brachytherapy
- 2019
-
Conference paper (peer-reviewed)abstract
- Purpose To develop a deep neural network for automatic reconstruction of needles in ultrasound images depicting the prostate during brachytherapy treatment of prostate cancer.Methods Ultrasound tomographies of the prostate from 907 treatments were used to train the artificial intelligent (AI) algorithm. The image matrices were downsampled to 128x128x128 and were used as in-data when training the AI, a 27 layer convolutional neural network. The needles were identified manually by medical physicists using conventional software. These reconstructions were used as gold standard when training the algorithm. An additional set of examinations were used for validation where the needle reconstructions by the AI were compared to the manual reconstructions. The root mean square deviation (RMSD) of needle position, including the central part (70 slices) of the needle was measured in order to avoid influence from artefacts around the needle tip. The result was also evaluated through visual inspection (see image). The times spent for manual vs. AI reconstruction were compared.Results RMSD for manual vs. AI reconstruction is on average (n=170) 1.18±1.0mm, whereas the difference between two manual operators is 0.02±0.06mm, which suggests that the AI is inferior to manual operators. The visual inspection, however, shows AI to be very accurate in positioning the needles. Manual reconstruction took approximately 11.0 minutes, whereas the time for the trained AI is negligible in comparison. Worth noticing regarding RMSD calculations is that, due to limited image resolution, small values may be under-estimated, hence overestimating the difference between the reconstruction methods.Conclusions The study implies that an AI may reconstruct needles for brachytherapy treatments of prostate cancer. The larger deviations between AI algorithm and manual operators, compared to between human operators appears to disagree with the high accuracy of the visual evaluation. However, visually, manual needle reconstructions appear to deviate more from the ultrasound image than do the AI reconstructions. This discrepancy is mainly caused by manual reconstruction software assuming straight needles, unlike the AI. We conclude that AI gives the opportunity to save a substantial amount of treatment planning time, when the patient is anesthetised. Further studies are needed to determine whether different reconstruction methods impact treatment plans.
|
|
3. |
- Högberg, Jonas, et al.
(author)
-
Comparison of Otsu and an adapted Chan-Vese method to determine thyroid active volume using Monte Carlo generated SPECT images
- 2024
-
In: EJNMMI Physics. - : Springer Nature. - 2197-7364. ; 11:1
-
Journal article (peer-reviewed)abstract
- BACKGROUND: The Otsu method and the Chan-Vese model are two methods proven to perform well in determining volumes of different organs and specific tissue fractions. This study aimed to compare the performance of the two methods regarding segmentation of active thyroid gland volumes, reflecting different clinical settings by varying the parameters: gland size, gland activity concentration, background activity concentration and gland activity concentration heterogeneity.METHODS: A computed tomography was performed on three playdough thyroid phantoms with volumes 20, 35 and 50 ml. The image data were separated into playdough and water based on Hounsfield values. Sixty single photon emission computed tomography (SPECT) projections were simulated by Monte Carlo method with isotope Technetium-99 m ([Formula: see text]Tc). Linear combinations of SPECT images were made, generating 12 different combinations of volume and background: each with both homogeneous thyroid activity concentration and three hotspots of different relative activity concentrations (48 SPECT images in total). The relative background levels chosen were 5 %, 10 %, 15 % and 20 % of the phantom activity concentration and the hotspot activities were 100 % (homogeneous case) 150 %, 200 % and 250 %. Poisson noise, (coefficient of variation of 0.8 at a 20 % background level, scattering excluded), was added before reconstruction was done with the Monte Carlo-based SPECT reconstruction algorithm Sahlgrenska Academy reconstruction code (SARec). Two different segmentation algorithms were applied: Otsu's threshold selection method and an adaptation of the Chan-Vese model for active contours without edges; the results were evaluated concerning relative volume, mean absolute error and standard deviation per thyroid volume, as well as dice similarity coefficient.RESULTS: Both methods segment the images well and deviate similarly from the true volumes. They seem to slightly overestimate small volumes and underestimate large ones. Different background levels affect the two methods similarly as well. However, the Chan-Vese model deviates less and paired t-testing showed significant difference between distributions of dice similarity coefficients (p-value [Formula: see text]).CONCLUSIONS: The investigations indicate that the Chan-Vese model performs better and is slightly more robust, while being more challenging to implement and use clinically. There is a trade-off between performance and user-friendliness.
|
|