REGULAR CONTENT
Final ID
448
Type
Original Scientific Research-Oral or Pos
Authors
S McLennan1, E Fayazzadeh2, R Gurajala3, S Shrikanthan4, F DiFilippo4, G McLennan5
Institutions
1Cleveland State University, Chagrin Falls, OH, 2N/A, Shaker Heights, OH, 3N/A, Beachwood, OH, 4Cleveland Clinic, Cleveland, OH, 5Cleveland Clinic, Chagrin Falls, OH
Purpose
New software developments to make tumor segmentation & radiation dose calculation have made it feasible to build dosimetry measurements into the normal workflow for Y-90 therapy. The process has become semi-automated in newer versions of the software. The purpose of this study was to validate the consistency of semi-automated software with prior manual segmentation methods.
Materials & Methods
In 9 random patients receiving Y-90, manual dosimetry was performed by drawing regions of interest around liver lobe and tumor margins & calculating radioactivity concentration (Bq/ml) of radiation deposited in each region after fusing pre-intervention MRs or CTs with post 90Y PETs. Activity was calculated using 68Ge as the reference radionuclide (1). The same images & regions were then used in a semi-automated workflow (MIM v. 6.01) using MIRD Kernel & Local Deposition Method (LDP) procedures. Dose to tumor and liver were compared between the two methods using Pearson correlation & Bland-Altman limits of disagreement. Measurement were equal if within 2 SD of the mean difference.
Results
Measurements were nearly identical and are reported in the table as mean dose ± standard deviation. The standard deviation is due to variability in tumor burden. Pearson coefficient was 0.9993 & 0.9985 for right tumor & lobe; 1.000 & 0.9995 for left tumor & lobe. Based on Bland-Altman analysis, the semi-automated method underestimated dose by 3.45, 1.82, 5.57, & 2.83 Gy for Rt tumor, Rt lobe, Lt tumor, & Lt lobe respectively. All measurements were were Within 2 standard deviations from the mean difference, consistent with statistical equivalence.
Conclusions
The semi-automated analysis software provides radiation dosimetry in Gy that correlates well with previously described methods.
Final ID
448
Type
Original Scientific Research-Oral or Pos
Authors
S McLennan1, E Fayazzadeh2, R Gurajala3, S Shrikanthan4, F DiFilippo4, G McLennan5
Institutions
1Cleveland State University, Chagrin Falls, OH, 2N/A, Shaker Heights, OH, 3N/A, Beachwood, OH, 4Cleveland Clinic, Cleveland, OH, 5Cleveland Clinic, Chagrin Falls, OH
Purpose
New software developments to make tumor segmentation & radiation dose calculation have made it feasible to build dosimetry measurements into the normal workflow for Y-90 therapy. The process has become semi-automated in newer versions of the software. The purpose of this study was to validate the consistency of semi-automated software with prior manual segmentation methods.
Materials & Methods
In 9 random patients receiving Y-90, manual dosimetry was performed by drawing regions of interest around liver lobe and tumor margins & calculating radioactivity concentration (Bq/ml) of radiation deposited in each region after fusing pre-intervention MRs or CTs with post 90Y PETs. Activity was calculated using 68Ge as the reference radionuclide (1). The same images & regions were then used in a semi-automated workflow (MIM v. 6.01) using MIRD Kernel & Local Deposition Method (LDP) procedures. Dose to tumor and liver were compared between the two methods using Pearson correlation & Bland-Altman limits of disagreement. Measurement were equal if within 2 SD of the mean difference.
Results
Measurements were nearly identical and are reported in the table as mean dose ± standard deviation. The standard deviation is due to variability in tumor burden. Pearson coefficient was 0.9993 & 0.9985 for right tumor & lobe; 1.000 & 0.9995 for left tumor & lobe. Based on Bland-Altman analysis, the semi-automated method underestimated dose by 3.45, 1.82, 5.57, & 2.83 Gy for Rt tumor, Rt lobe, Lt tumor, & Lt lobe respectively. All measurements were were Within 2 standard deviations from the mean difference, consistent with statistical equivalence.
Conclusions
The semi-automated analysis software provides radiation dosimetry in Gy that correlates well with previously described methods.