Development of New Imaging Biomarkers for Quantitative Evaluation of Prognosis and Treatment Response in Prostate Cancer Patients

Elin Trägårdh, Anders Bjartell, Fredrik Kahl, Lars Edenbrandt

Background and aim: Bone is a frequent site of metastases in prostate cancer, and the presence of bone metastases is associated with poor outcome. Bone Scan Index (BSI) obtained from bone scintigraphy is a method of expressing the tumour burden in bone as a percentage of the total skeletal mass. We have developed an automated method for calculating BSI. We will continue to validate BSI in patients with prostate cancer. We will also start to develop positron emission tomography – computed tomography (PET-CT) imaging biomarkers for scans that visualise bone metastases.

Plan: For BSI studies, we will perform a study on reproducibility, as well as retrospective and prospective studies. Studies will be performed in collaboration with Gothenburg and Odense. For PET-CT studies, we will first develop the method by segmentation of the skeleton from the CT images and then calculate “3-dimensional” BSI from the PET-images. Validation of the new method will be performed in collaboration with Gothenburg, Odense and Munich. This is a PhD project with a time plan 2015-2021.

Significance: New imaging biomarkers that provide relevant and incremental diagnostic information may be an important step towards personalised medicine. If an imaging biomarker can be used to select the patients who will benefit from a specific drug or combination of drugs, then this will be of advantage to patients. Side effects can be avoided in patients who would not benefit from the treatment.



Smoker's lung disease: an integrated approach to pathophysiology and diagnosis using lung function measurements, plasma biomarkers and imaging

Per Wollmer, Lars Edenbrandt, Sven Valind

Smoking is a major cause of morbidity and mortality. One fifth of smokers develop chronic obstructive lung disease (COPD). COPD is not restricted to the lungs but has extra-pulmonary manifestations, many of which are thought to be mediated by systemic inflammation. Smoking is also the major cause of lung cancer. The relation between lung function, local and systemic inflammation is not known. Positron emission tomography (PET/CT) opens new possibilities to measure local inflammation and to diagnose lung cancer by using fluorodeoxyglucose (FDG), a tracer taken up avidly by both activated leukocytes and cancer cells.

One of the aims with this project is to develop quantification and computer-aided diagnosis (CAD) methods for imaging studies of inflammation in the lungs as well as evaluation of solitary pulmonary nodules. Our preliminary observations indicate that lung uptake of FDG is increased in smokers relative to non-smokers, reflecting local inflammation. We will therefore measure FDG uptake in lung tissue prospectively in subjects undergoing PET/CT and relate local inflammation to lung function and indicators of systemic inflammation. PET/CT and FDG is frequently used for evaluation of lung cancer. Interpretation is complex and hampered by the low and variable density of the lung. We plan to develop an automated system for image analysis and decision support based on machine learning tools.


Clinical Decision Support Tools in Cardiac Imaging

Elin Trägårdh

Stress myocardial perfusion scintigraphy (MPS) is widely regarded as a clinically useful non-invasive imaging modality for diagnosing patients with suspected coronary artery disease. However, localized soft-tissue attenuation by the breasts, lateral chest wall, and abdomen may create artefacts that mimic true perfusion abnormalities and decrease test specificity. The American Society of Nuclear Cardiology and the Society of Nuclear Medicine conclude in their joint statement from 2004 that incorporation of attenuation corrected (AC) images in addition to ECG gating will improve image quality, interpretive certainty, and diagnostic accuracy. These results are anticipated to have a substantial impact on improving the effectiveness of care and lowering health care costs. Commonly used MPS software, however, usually only include normal stress database for non-attenuation corrected (NC) images.

The aim of this project is to develop and compare different normal stress databases for MPS, with regard to men, women, AC and NC images. We also want to determine to many subjects are needed in a normal stress database. Finally, we want to compare the accuracy of computer interpretation of myocardial infarction when using a NC normal stress database vs. and AC normal stress database. We have access to all MPS studies performed in Malmö 2004-2008.

In this project, we collaborate with Odense University Hospital in Denmark, the Sahlgrenska University Hospital in Gothenburg, Sweden, and EXINI Diagnostics AB, Lund, Sweden.

The research is supported by grants from ALF.


Quantification and Computer-Aided Diagnosis in Diagnostic Imaging

Lars Edenbrandt

Reading and reporting diagnostic images is a difficult process. False image interpretations may have serious consequences, for example patients may undergo unnecessary treatment or be denied an essential treatment. The objectives of this project are to develop and evaluate
quantification and computer-aided diagnosis (CAD) systems regarding the interpretation of diagnostic PET/CT and bone scan images in patients with cancer. These systems are intended to help physicians read and report the images with high quality.

We have previously developed quantification and CAD systems that have shown to be of benefit for the physicians interpreting diagnostic images. We will now adjust our systems to meet requirements in different clinical settings and to perform clinical evaluations of our
systems. Modern image processing and machine learning techniques will be developed for the image analysis. These techniques "learn" from examples and we will use our large databases of PET/CT and bone scan images. The clinical evaluation of our systems will be performed
together with a research group generating information to be used in diagnosis, prognostic evaluation, and follow-up of patients with prostate cancer in collaboration with Memorial Sloan Kettering Cancer Center, New York.

CAD systems based on our previous research results are now used clinically in 15 countries. The results obtained in this project will be of great value for future CAD systems.

This project is accomplished in close collaboration with medical research groups at the Skåne University Hospital in Malmö, the Sahlgrenska University Hospital in Gothenburg, the PET Center, Odense University Hospital in Denmark, and the Memorial Sloan-Kettering Cancer Center, USA. Technical partners includes researchers from the Mathematical Imaging Group and the Computational Biology & Biological Physics Group at Lund University, Sweden as well as EXINI Diagnostics AB, Lund Sweden.

The research is supported by grants from the Swedish Research Council,
ALF, Region Skåne, and Lundgrens stiftelse.


Optimisation and comparison of different reconstruction algorithms

Marcus Söderberg, Sören Mattsson, Sven Valind, Helena Uusijärvi-Lizana, Sigrid Leide-Svegborn

Historically, filtered back projection has been the main method of reconstruction used in SPECT, but in recent years more and more being replaced by iterative algorithms due to today’s available computational power and possibility to include compensation for image-degrading effects. New developed reconstruction algorithms and noise reduction methods needs to be optimised and clinically evaluated.

In this project we will optimise Siemens iterative algorithm Flash 3D in terms of equivalent number of iterations, defined as the number of subsets times the number of iterations for 123I –MIBG examinations. The reconstructions will be compared with two recently developed reconstruction algorithms ReSPECT (iterative) and OPED (analytic), made available in the frame of the MADEIRA project (see below). The optimisation and comparison will be visually assessed by nuclear medicine physicians, using the software package Scientific Visualizer platform (Scivis wissenschaftliche Bildverarbeitung, Göttingen, Germany).

In this project, we collaborate with Scivis wissenschaftliche Bildverarbeitung in Göttingen, Germany and Helmholtz Zentrum in Munich, Germany.

The research is carried out within the Collaborative Project “MADEIRA” (, cofunded by the European Commission through EURATOM Seventh Framework Programme (Grant Agreement FP7-212100).


The Madeira phantom

Marcus Söderberg, Sören Mattsson, Sigrid Leide-Svegborn

Essentially for detection of  lesions in SPECT and PET images is the contrast to noise ratio (CNR) of lesion to background. As there is always an intrinsic resolution volume defined by camera, collimator, radionuclide, acquisition protocol and reconstruction method, the size of the lesion is also important. Below a certain volume, the reconstructed intensity tends to diffuse into neighbouring voxels, thereby giving too low target to background ratios. This effect is called partial volume effect (PVE). To investigate the importance of this effect and to find methods to correct for it, a new patent filed phantom (European patent application no. 09008184, “Phantom for a tomographic medical imaging apparatus”), the “MADEIRA” phantom, has been designed so that it simultaneously can provide different target to background ratios with a linearly changing diameter of active or inactive lesions.

The aim of the project is to characterize different nuclear medicine tomographic systems and reconstruction algorithms in their performance and behaviour concerning PVE and detectability by varying the acquisition parameters and the count statistics.

In this project, we collaborate with Scivis wissenschaftliche Bildverarbeitung in Göttingen, Germany.

The research is carried out within the Collaborative Project “MADEIRA” (, cofunded by the European Commission through EURATOM Seventh Framework Programme (Grant Agreement FP7-212100).


Biokenetics and dosimetry of DaTSCAN™ ioflupane (123I) in patients with suspected Alzhaimer's desease or Parkinsonism

Marie Sydoff, Sigrid Leide-Svegborn, Sören Mattsson, Sven Valind

The use of 123I-ioflupane for the investigation of Alzheimer´s disease, Parkinsonism or Lewy body dementia is today common in the nuclear medicine clinical departments. This calls for the need of determining the biokinetics and activity content in the human body in order to estimate the absorbed doses and the effective dose to patients undergoing this kind of examinations and to optimize the time for the clinical measurements.

In this study, the biokinetics and dosimetry of DaTSCAN™ ioflupane (123I) is investigated in patients with suspected Alzheimer’s disease or Parkinsonism. Anterior and posterior wholebody scans are performed on the subjects at five different time points post injection (p.i.) and SPECT/CT thoracic and brain images are also acquired. Additionally, blood and urine samples are collected during the whole measurement period. These images and samples are used in order to quantify the activity content in several organs at different time points and thus to estimate the uptake and excretion of the radiopharmaceutical over time, which is needed in order to determine the absorbed and effective dose to the patients.

The research is supported by grants from the CEC 7th framework program MADEIRA and from the Swedish Radiation Safety Authority.


Absolute quantification of activity content from PET images using the Philips Gemini TF PET/CT system

Marie Sydoff, Sigrid Leide-Svegborn, Sören Mattsson

Positron emission tomography combined with computed tomography (PET/CT) is a quantitative technique suitable for diagnostics and uptake measurements. The quantitative results can be used for the purpose of calculating the absorbed dose to patients undergoing nuclear medicine investigations. Hence, the accuracy of the quantification of the activity content in organs or tissues is of great importance.

When using a Philips Gemini TF PET/CT system as the one that we have at the clinic, the activity concentration in a region of interest (ROI) is given by the system. The reliability of the activity concentration values given by the system is studied using a Jaszczak phantom containing hot spheres of different sizes; the influence of the ROI size and the impact of organ size, the so-called partial volume effect, are investigated with different lesion-to-background ratios in the phantom. The influence of the partial volume effect is important to determine, not only for dose assessments, but since it can also pose problems when using PET images for radiotherapy treatment planning, seeing that the tumours would look less aggressive than they really are.

The research in supported by grants from the CEC 7th framework program MADEIRA.


Online Support for Viewing, Interpreting and Reporting DMSA-scans

Per Wollmer, Lars Edenbrandt, Mikkel B. Stegmann, Karl Sjöstrand

Children with repeated urinary tract infections may suffer from reduced kidney function. To quantify the extent and localization of the reduction, a nuclear image known as a DMSA-scan is obtained. The DMSA tracer compound accumulates in the kidneys with reduced accumulation in damaged regions, making it possible to quantify the severeness of the condition. The investigation is performed relatively rarely; thus there are few expert clinicians to interpret the images. Further, the methods used to delineate the kidneys and damaged kidney regions, and the structure of the resulting reports differ between hospitals.

The nuclear medicine group in Malmö are taking part in a nation-wide project coined the "Image Laboratory" - the development of an online "app-store"-like service through which research groups and small companies can reach a large number of potential customers, and clinicians can browse and access the latest developments in medical imaging research. Our contribution to this project is to develop an online application for viewing, interpreting and reporting DMSA-scan images. This will enable clinicians to process pending investigations anywhere, at any time, using their office or home computer, a tablet, or a mobile phone.

The research in supported by grants from Vinnova.