SPREAD-project

Software Performance and Reproducibility of Emphysema Assessment: Demonstration

5th Framework Programme of the European Union QLG1-CT-2000-01752

Full title:
Demonstration of the Performance of a New Analytical Software Package Developed for Detection of Progress of Emphysema by CT

January 2001 - September 2004 SPREAD

Background

In 1996, the WHO recommended that the rare hereditary disorder Alpha-1 Antitrypsin Deficiency (AAD) should be better characterized in the population in the EU and other parts of the world. The recommendation was needed because of the severe diseases that can be caused at young age by this disorder. Out of three different diseases, the lung disease emphysema is an important one. In 1997, a working party of the European Respiratory Society established a registry in the EU for subjects with AAD. This registry greatly facilitates the recruitment for studies such as the current study.
As mentioned above, emphysema occurs often in subjects with AAD. Emphysema is a disease in which the oxygen exchange area in the lungs is destroyed by enzymes. AAD subjects have very low blood levels of the important enzyme inhibitor alpha-1-antitrypsin. Therefore, they develop emphysema at young age. The progress of emphysema is currently measured by lung function tests, which require several years to detect small changes. New, more sensitive methods to detect progress of emphysema are needed for assessment of prognosis and for clinical trials with new drugs. In contrast to normal subjects with emphysema, those with AAD will develop emphysema in early adult life, and with rapid progress. Therefore, this is an ideal population to study new, more sensitive methods to detect the progress of emphysema.

PulmoCMSThe introduction of Computer Tomography (CT) in the field of medical imaging was a major s/tep in diagnosing emphysema. Several cross-sectional studies, published in the literature on quantitative aspects of CT images of the lung, demonstrated the possibilities of computerized detection of emphysema and quantitative scoring. A common approach is to detect the lung boundaries in the CT data and to calculate the frequency distributions of densities, expressed in Hounsfield units, within the identified lung regions. Significant differences in parameters derived from these frequency distributions have been found to distinguish lungs with emphysema from normal lungs or lungs with other diseases. A recent longitudinal study of emphysema in AAD subjects showed that CT densitometry of the lung, using our newly developed analytical software (PulmoCMS), was 2.5 fold more sensitive to show the progress of emphysema compared with lung function tests. A study on the assessment of Lung Volume Reduction Surgery (LVRS) has shown that changes in the CT-parameters due to LVRS correlate very well with functional changes measured by Total Lung Capacity (TLC) and Residual Lung Volume (RV).

Objectives

The objectives of this study is to demonstrate that this previously obtained result with our software approach can be reproduced on a larger scale in a multi-centre study using different brands of equipment. Reproduction of this result on a larger scale would be a major scientific advancement in the R&D area of this disease. We would like to demonstrate to the drug registration authorities in Brussels and the pharmaceutical industry that our software approach is a powerful tool for the assessment of emphysema and other lung diseases. It has great potential to replace pulmonary function tests as a measurement tool for future clinical drug trials because it may prove drugs effects much earlier.
The scientific and technological objectives are:

  1. To show that the use of CT scanners from five different market leading CT vendors does not increase the variability in the measurements of these data;
  2. To show that data analyses by different software operators result in a comparable outcome as demonstrated by the single operator who developed the software.

Partners

Partner 1 - Coordinator

Leiden University Medical Center

Netherlands

Prof. J.H.C. Reiber

J. Stolk, MD, pulmonologist

E.L. van Persijn van Meerten, MD, radiologist

B.C. Stoel, PhD

M.E. Bakker, PhD

 LUMC

Partner 2

Gentofte Hospital - Copenhagen

Denmark

Prof. A. Dirksen, pulmonologist

U. Sander

S.B. Shaker, PhD Student

 Gentofte

Partner 3

Queen Elizabeth Hospital

University of Birmingham

United Kingdom

Prof. R.A. Stockley, pulmonologist

P. Guest, MD, radiologist

D. Parr, PhD

P. Dawkins, PhD

 Birmingham

Partner 4

Malmö University Hospital

Lunds University

Sweden

E. Piitulainen, MD, pulmonologist

B. Hillarp, MD, radiologist

 Malmo

Partner 5

University Hospital Zürich

Switzerland

Prof. E. Russi, pulmonologist

Th. Boehm, MD, radiologist

E. Grebski, MD

 Zurich

CT Scanners The overall aim of this study is to demonstrate that previously obtained results with the PULMO-CMS software approach can be reproduced on a larger scale in a multi-centre study using different brands of equipment. Each partner in this project has a different CT scanner.

Partner 1 - Coordinator

Leiden University Medical Center

Netherlands

Toshiba Aquilion

Multi Slice CT scanner

Partner 2

Gentofte Hospital - Copenhagen

Denmark

General Electric LightSpeed

Multi Slice CT scanner

Partner 3

University of Birmingham

United Kingdom

Philips MX 8000

Multi Slice CT scanner

Partner 4

Lunds University

Sweden

Philips Tomoscan

Spiral CT scanner

Partner 5

University Hospital Zurich

Switzerland

Siemens VolumeZoom

Multi Slice CT scanner

Acquisition Protocol

Water Calibration

To ensure the constancy of the CT density measurement, the scanner should be calibrated for water on the same day of the image acquisition. This calibration procedure is needed since even a small drift of a few Hounsfield Units (HU) may affect the measured loss of lung tissue. The calibration procedure will be done prior to the scan of the first patient and should be specific for the acquisition protocol with the use of the proper water phantom.

Air Calibration

The air calibration will be done according to the manufacturer's guidelines and instructions. The first air calibration is to be performed within three hours before the CT acquisition of the first patient, and must be repeated at least once every three hours

Quality control To check the performance of the scanner, a dedicated foam phantom will be scanned on the same day when patients are scanned. The foam phantom is scanned using the protocol while placed on the table.

Perspex box Compartments containing 15 pieces of polyethene foam Densities ranging from 15 g/l to 65 g/l

 Phantom

CT-Acquisition

All subjects will be scanned twice in the supine position by spiral (single- or multi-slice) CT of the chest, both during full inspiration. Following 3 deep breaths, the patient will be asked to take another deep breath and to hold it for the duration of the scan.
The scan will be taken from the diaphragm in the direction of the neck, to prevent inhalation artifacts in the images. The scanning parameters are dependent on the type of scanner used but finally will result in reconstructed images of 5 mm slice thickness with an increment (overlap) of 2.5 mm.

Scanning Parameters

Toshiba
Toshiba Aquilion
(4 detectors)

kVp

135

mA

40 (20 mAs per rotation)

Table feed

30 mm (detector pitch 6; beam pitch 1.5)

Rotation time

0.5 s

Collimation

4x5 mm

Field

L (400mm)

Field of View (FOV)

equal to the FOV of the first scan

Reconstruction

5 mm slice thickness, 2.5 mm increment (overlap)

Reconstruction filter

FC12

General Electric
General Electric LightSpeed
(4 detectors)

kVp

140

mA

40 (32 mAs per rotation)

Table feed

30 mm (detector pitch 6; beam pitch 1.5)

Rotation time

0.8 s

Collimation

4x5 mm

Field of View (FOV)

equal to the FOV of the first scan

Reconstruction

5 mm slice thickness, 2.5 mm increment (overlap)

Reconstruction filter

soft

Philips MX 8000
Philips MX 8000
(4 detectors)

kVp

140

mA

40 (20 mAs per rotation)

Table feed

30 mm (detector pitch 6; beam pitch 1.5)

Rotation time

0.5 s

Collimation

4x5 mm

Field of View (FOV)

equal to the FOV of the first scan

Reconstruction

5 mm slice thickness, 2.5 mm increment (overlap)

Reconstruction filter

A

Philips AVEU
Philips Tomoscan AVEU
(1 detector)

kVp

140

mA

25 (25 mAs per rotation)

Table feed

14 mm (beam pitch = detector pitch = 2)

Rotation time

1 s

Collimation

7 mm

Field of View (FOV)

equal to the FOV of the first scan

Reconstruction

5 mm slice thickness, 2.5 mm increment (overlap)

Reconstruction filter

1

Siemens Volume Zoom
Siemens Volume Zoom
(4 detectors)

kVp

140

mA*

60 (30 mAs per rotation)

Table feed

30 mm (detector pitch 6; beam pitch 1.5)

Rotation time

0.5 s

Collimation

4x5 mm

Field of View (FOV)

equal to the FOV of the first scan

Reconstruction

6 mm slice thickness, 3 mm increment (overlap)

Reconstruction filter

BF 10 (very smooth)

* Warning: Siemens has introduced an alternative parameter to set the mA setting of the CT scanner, accounting for the pitch. The so-called effective mAs is defined as mAs x 4 / detector pitch. The lowest possible setting for the eff. mAs is 20. The mA setting can be derived from the effective mAs:
mA = (Eff. mAs / Rotation time) x (det.pitch/4).

Workplan

Each partner will randomly select 23 patients from the EU registry of alpha-1-antitrypsin deficiency, amounting to a total of 115 patients (see diagram 1).

At baseline and after 29 months, all subjects will be scanned by spiral (or multislice) CT of the chest and their function will be measured.

Each partner will use Pulmo-CMS to derive the densitometric parameters from the CT images.

Additionally, all image data will be sent to the coordinator for re-analysis.

In a statistical analysis, a comparison will be made between the sensitivity of CT densitometry and of lung function tests in assessing the progress of emphysema.

The inter- and intra-observer variability of the CT measurements will be determined (see diagram 2).

The results will be reported to the Committee for Proprietary Medicinal Product (CPMP) of the European Medicines Evaluation Agency (EMEA) in Brussels.

Diagram1 

Diagram 1 Each partner will select 23 patients. From each patient CT scans and lung function tests will be performed at baseline and follow-up.

Each partner (incl. coordinator):

  • Recruitment of 23 patients with AAD
  • CT scan acquisition (2x) and lung function test at baseline
  • Analysis CT images baseline with Pulmo-CMS
  • Sent all baseline results to coordinator
  • CT scan acquisition (2x) and lung function test at follow-up
  • Analysis CT images follow-up with Pulmo-CMS
  • Sent all follow-up results to coordinator

Coordinator:

  • Re-analysis of CT images baseline by all partners using Pulmo-CMS
  • Determination of inter-observer variability in baseline measurements
  • Re-analysis of CT images baseline by partner 1 with Pulmo-CMS
  • Determination of intra-observer variability in baseline measurements of partner 1
  • Re-analysis of CT images follow-up by all partners using Pulmo-CMS
  • Determination of inter-observer variability in follow-up measurements
  • Determination of overall inter-observer variability in all measurements
  • Comparison of progress of emphysema (baseline -> follow-up) based on CT and lung function based on statistical analysis

Diagram2 

Diagram 2 Each partner will perform CT scans and lung function tests at baseline and follow-up. Each partner will analyze the CT images with the PULMO-CMS software. The CT images and the analysis results together with the lung function test results will be sent to the coordinator who will perform a second analysis of the CT images for the determination of the inter-observer and intra-observer variability. The coordinator also will statistically analyze the baseline and follow-up CT data and lung function test data in order to determine the progress of emphysema obtained by the two methods. Finally, both methods will be compared and evaluated for their sensitivity to detect progression of emphysema

 

Contact

LUMC -
Coordinator

J.H.C.Reiber@lumc.nl

B.C.Stoel@lumc.nl

J.Stolk.long@lumc.nl

EPersijn@lumc.nl

M.E.Bakker@lumc.nl

For general information, please contact:


M.E. Bakker, PhD.
Division of Image Processing
Department of Radiology, 1-C2S
Leiden University Medical Center
P.O. Box 9600
2300 RC Leiden
The Netherlands
Tel. +31 (0)71 526 1246
Fax. +31 (0)71 526 6801
e-mail: M.E.Bakker@lumc.nl