s demehri 1, m.k kalra 2, m.l steigner 1, f.j rybicki 1, m.j. lang, 3, s.g silverman 1. 1.department...

20
S Demehri S Demehri 1 , M.K Kalra , M.K Kalra 2 , M.L Steigner , M.L Steigner 1 , F.J , F.J Rybicki Rybicki 1 , M.J. Lang, , M.J. Lang, 3 , S.G Silverman , S.G Silverman 1 . . 1. 1. Department of Radiology, Brigham & Women's Department of Radiology, Brigham & Women's Hospital , Harvard Medical School Hospital , Harvard Medical School 2. 2. Department of Radiology, Massachusetts General Department of Radiology, Massachusetts General Hospital, Harvard Medical School Hospital, Harvard Medical School 3. 3. Department of Biological Engineering, Department of Biological Engineering, Massachusetts Institute of Technology Massachusetts Institute of Technology 4. 4. Presented Presented RSNA 2009 RSNA 2009 Novel CT-based Method for Determining Urinary Stone Volume: Phantom and Clinical Studies

Upload: darlene-weaver

Post on 03-Jan-2016

216 views

Category:

Documents


2 download

TRANSCRIPT

S Demehri S Demehri 11, M.K Kalra , M.K Kalra 22, M.L Steigner , M.L Steigner 11, F.J , F.J Rybicki Rybicki 11, M.J. Lang,, M.J. Lang, 33, S.G Silverman , S.G Silverman 11. .

1.1. Department of Radiology, Brigham & Women's Hospital , Department of Radiology, Brigham & Women's Hospital , Harvard Medical SchoolHarvard Medical School

2.2. Department of Radiology, Massachusetts General Hospital, Department of Radiology, Massachusetts General Hospital, Harvard Medical School Harvard Medical School

3.3. Department of Biological Engineering, Massachusetts Institute Department of Biological Engineering, Massachusetts Institute of Technology of Technology

4.4. Presented RSNA 2009Presented RSNA 2009

Novel CT-based Method for Determining Urinary Stone Volume: Phantom and

Clinical Studies

Why Stone Volume?• Urolithiasis : chronic and largely recurrent diseaseUrolithiasis : chronic and largely recurrent disease• CT used to depict size, number, and CT used to depict size, number, and location of urinary

calculi to assess metabolic activity:

• Stone growth or new stone Stone growth or new stone formation in one year formation in one year

•Stone size has been typically described Stone size has been typically described using maximal axial diameter, which is a using maximal axial diameter, which is a poor indicator of stone volumepoor indicator of stone volume

Attenuation Threshold-based CT Methods

• Attenuation threshold-based CT methods can quantify Attenuation threshold-based CT methods can quantify volume of calcium in the coronary arteries : Coronary calcium score

• Such methods could be applied both at diagnosis and Such methods could be applied both at diagnosis and follow-up of urinary calculi to determine if medical follow-up of urinary calculi to determine if medical therapy is necessary, and to monitor the success of therapy is necessary, and to monitor the success of medical treatment. medical treatment.

• In this project, we described fixed (FTM) and variable (VTM) attenuation-threshold based methods for quantification of volume of urinary stones

Material & Methods:Gold Standard for Stone Volume

• Thirty six calcium oxalate stones were obtained Thirty six calcium oxalate stones were obtained from a human stone bank (de-identified from a human stone bank (de-identified samples). samples).

• Gold standard for stone volumes : Gold standard for stone volumes : Fluid-displacement technique

• With repeated measurements of a flask mass With repeated measurements of a flask mass filled with double-distilled water, the standard filled with double-distilled water, the standard error of this method was estimated to be 4 mmerror of this method was estimated to be 4 mm33. .

Material & Methods:Phantom Experiment

• Stones embedded in ground Stones embedded in ground meat (approximately 30 HU) meat (approximately 30 HU) and placed in dosimetric holes and placed in dosimetric holes in abdominal slices of in abdominal slices of Alderson-Rando Phantom.Alderson-Rando Phantom.

• Phantom was scanned using a Phantom was scanned using a CT protocol (120 kVp, 200 CT protocol (120 kVp, 200 mAs) in our clinical practice mAs) in our clinical practice for patients with acute flank for patients with acute flank painpain

Material & Methods:Phantom Experiment

• All images reconstructed at All images reconstructed at 0.6, 1, and 3 mm thick sections 0.6, 1, and 3 mm thick sections with no overlap. with no overlap.

• The phantom rescanned using The phantom rescanned using lower tube currents (150, 100, lower tube currents (150, 100, and 50 mAs; other parameters and 50 mAs; other parameters held constant) and tube held constant) and tube potentials (80 and 100 kVp potentials (80 and 100 kVp maintaining a constant CTDI).maintaining a constant CTDI).

Material & Methods:Phantom Experiment

• Open source software (OsiriX, Mac OS X Version Open source software (OsiriX, Mac OS X Version 3.3.2) used to segment the urinary stones. 3.3.2) used to segment the urinary stones.

• Using Using FTMFTM, stones were segmented using two fixed , stones were segmented using two fixed attenuation thresholds: attenuation thresholds:

1.1. 130 HU (the threshold used to determine coronary 130 HU (the threshold used to determine coronary calcium volume) calcium volume)

2.2. 575 HU i.e., attenuation value equal to half of the mean 575 HU i.e., attenuation value equal to half of the mean attenuation of all stones (1150+/-290, mean+/-SD).attenuation of all stones (1150+/-290, mean+/-SD).

• Using Using VTMVTM, stones segmented using individual , stones segmented using individual attenuation-thresholds equal to half of the attenuation attenuation-thresholds equal to half of the attenuation of each stone.of each stone.

Material & Methods:Clinical Study-Interobserver Agreement

• A retrospective study in 17 patients (M/F=1.6, A retrospective study in 17 patients (M/F=1.6, mean age=55) after IRB-approval with urinary mean age=55) after IRB-approval with urinary calculi. calculi.

• Two independent observers measured the Two independent observers measured the maximum axial diameter of urinary stones stone maximum axial diameter of urinary stones stone volumes using both the FTM and the VTM.volumes using both the FTM and the VTM.

Material & Methods:Statistical Analysis

• Normalized (%) Bland-Altman analysis used to Normalized (%) Bland-Altman analysis used to evaluate the mean difference (i.e., bias) and evaluate the mean difference (i.e., bias) and standard deviation of difference (i.e., precision) standard deviation of difference (i.e., precision) between each CT method and the gold standardbetween each CT method and the gold standard

• To examine the effect of CT parameters (section To examine the effect of CT parameters (section thickness, mAs, and kVp) on the accuracy and thickness, mAs, and kVp) on the accuracy and precision of the stone volume measurements: one-precision of the stone volume measurements: one-way analysis of variance (ANOVA).way analysis of variance (ANOVA).

• To estimate interobserver agreement: Bland-Altman To estimate interobserver agreement: Bland-Altman limits of agreement for the paired measurements limits of agreement for the paired measurements for each method.for each method.

Normalized Bland-Altman Difference Plot CT–derived Stone Volume Measurement Using FTM,

Threshold Level at 130 HU

Measurements compared with the volume measurement using Measurements compared with the volume measurement using fluid displacement technique, the fluid displacement technique, the gold standardgold standard. .

FTM with Threshold Level at 575 HU- Half of the Mean Attenuation of All Stones

VTM- Threshold Level Set at the Half of the Attenuation of Each Stone

aa bb

cc

Fixed (FTM) and Variable (VTM) Fixed (FTM) and Variable (VTM) threshold methods in phantom threshold methods in phantom study. Using FTM, with study. Using FTM, with segmentation threshold is set at 130 segmentation threshold is set at 130 HU (as in coronary calcium score) HU (as in coronary calcium score) (a) and at 575 HU (b), respectively. (a) and at 575 HU (b), respectively. Using VTM (c) with segmentation Using VTM (c) with segmentation threshold set at 699 HU (half of the threshold set at 699 HU (half of the stone attenuation (1398 HU)). The stone attenuation (1398 HU)). The gold standard stone volume was gold standard stone volume was 119.9 mm119.9 mm33..

220.6 mm220.6 mm33 134.5 mm134.5 mm33

114.9 mm114.9 mm33

Stone volume measurements using VTM as well as mean stone attenuations at various section widths, tubecurrents, and potentials were compared to the volume measurement (†) with CT protocol used in our clinicalpractice.

Stone attenuation (HU)

Mean +/- SD

Stone volume (mm3)Mean +/- SD

Bias (%) Precision(%)

VTM at variable section widths (200 mAs, 120 kVp)

0.6 mm† 1150+/-290 55.1+/-51.2 2 20

1 mm 1140+/-280 54.3+/-49.3 5 28

3 mm 860+/-330 *** 49.7+/-49.2 26 * 87 ****

VTM at variable mAs (120 kVp, 0.6 mm section width)

150 mAs 1130+/-290 54.6+/-50.6 2 22

100 mAs 1150+/-300 52.8+/-50 -1 22

50 mAs 1125+/- 275 52.2+/-50 - 2 22

VTM at variable kVp (with constant CTDI, 0.6 mm section width)

100 kVp 1280+/-380 *** 52.1+/-49.4 -4 18

80 kVp 1440+/-420 *** 52.7+/-51.8 -4 21

Normalized Bland-Altman analysis was carried out to assess the biases (i.e., mean percentage differences betweeneach CT method and reference standard) and precisions (i.e., SD of percentage differences) of CT-based volumemeasurements. (*, p< 0.05; ****, p<0.0001)

aabb

cc

The stone volume was The stone volume was measured using variable measured using variable threshold method (VTM) in 0.6 threshold method (VTM) in 0.6 mm (a), 1 mm (b), and 3 mm (c) mm (a), 1 mm (b), and 3 mm (c) section widths. The stone section widths. The stone volume using fluid volume using fluid displacement technique as gold displacement technique as gold standard was standard was 4.1 mm4.1 mm33..

5.5 mm5.5 mm33 4.9 mm4.9 mm33

9 mm9 mm33

Mean percentage difference and Bland-Altman 95% limits of agreement for inter-observer comparisons of stone maximum axial diameter and volume measurementsusing fixed (FTM) and variable (VTM) threshold methods in seventeen patients withurinary calculi.

Measurements Mean Percentage Difference between

Observers*

95% Limits of Agreement (%)

Correlation Coefficient (r)*

Volume using FTM

8.7 (0.7, 16.7) -25, 42.4 >0.99(0.997-0.999)

Volume using VTM

5.3 (-3.3, 13.8) -32.4, 45 0.974(0.949-0.99)

Maximum AxialDiameter

12.4 (7.9 , 16.9) -9, 34.4 0.98(0.953-0991)

* Numbers in parentheses are the 95% confidence interval.

Study Limitations

• Phantom study was limited to calcium oxalate Phantom study was limited to calcium oxalate stonesstones

• Imaging of stone in phantom is an idealized clinical Imaging of stone in phantom is an idealized clinical scenario ie., no motion, no radio-opaque scenario ie., no motion, no radio-opaque surrounding structure (e.g., nephrostomy tube)surrounding structure (e.g., nephrostomy tube)

• In phantom study, the observer was blinded to gold In phantom study, the observer was blinded to gold standard stone volume but not the CT parameters.standard stone volume but not the CT parameters.

• The gold standard, has measurement error which The gold standard, has measurement error which could affect the bias and precision of CT based could affect the bias and precision of CT based measurements.measurements.

Novel CT-based Stone Volume MeasurementNovel CT-based Stone Volume Measurement

• Attenuation threshold-based CT method can accurately and Attenuation threshold-based CT method can accurately and precisely quantify urinary stone volumesprecisely quantify urinary stone volumes

• Individually-derived attenuation-thresholds are superior to fixed Individually-derived attenuation-thresholds are superior to fixed thresholdsthresholds

• Thinner collimations yield more accurate volume measurementThinner collimations yield more accurate volume measurement

• Measurements can be performed using low radiation dose CT Measurements can be performed using low radiation dose CT protocolsprotocols

“The sum of each stone’s volume could be used to derive a cumulative stone volume or “stone score” (akin to coronary calcium score), which may help the radiologist obtain an accurate and precise method of quantifying individual as well as cumulative stone burden and metabolic activity. “

Thank youThank you