Molecular Imaging Professor Sir Michael Brady FRS FREng Department of Engineering Science Oxford University
Over the past 20 years, we have developed new ways to image anatomy, new ways to see inside the body, noninvasively We can watch the body in action, as it responds to the injection of a drug or contrast agent, to highlight aberrant physiology We can watch the body functioning in a whole range of ways the brain thinking, degradation in white matter, and the pulsing of the heart Now we are beginning to image cellular and molecular processes the convergence of molecular biology and image analysis
Convergence of molecular medicine and image analysis Molecular medicine The genetic basis of disease new therapies and new understanding, eg stem cells Modelling cellular and molecular processes Modelling drug interactions at the cellular level Pharmacokinetic processes Medical image analysis Image segmentation, fusion, image alignment, and modelling change Major changes have been taking place in image analysis too
Molecular Imaging Dream: from treatment of symptom to understanding of cause to anticipation of disease - personalised medicine Aspiration: integrated systems biology Transform drug use Transform drug discovery Colorectal cancer Personalised optimal drug doses Preclinical image analysis Context: convergence of molecular medicine and (medical) image analysis Current reality: measurement of disease processes, primarily using nuclear medicine: PET & SPECT
For every 10,000 target compounds, one makes it to market It costs $100M-$1Bn to bring a drug to market every month saved $1M to bottom line Most patents are now filed by Biotech companies Major pharmaceutical companies are looking for novel processes for drug discovery, personalising dose determination, and monitoring response & disease progression Molecular Imaging is a promising candidate
The left column shows the image pre-treatment, the centre column indicates initial response to treatment, while the third shows that there has been regression
Images & biology PET (and Spect) PK modelling (MRI, PET, Spect) Quantitative analyses Model-based segmentation of tumours Dynamic PET reconstruction & analysis Parametric maps Angiogenesis, glycolysis Hypoxia (HIF), cell division (thymidine), ph control Dopaminergic activity (Alzheimer s) Caudate/putamen degeneration (Parkinson s)
Model-based PET segmentation PET image noise model (inside and outside tumour) + Pharmacokinetic model of 18 FDG take up FDG in plasma (q1) K 1 k 2 FDG in tissue (q2) (Glucose transporter) k 3 k 4 FDG-6P in tissue (q2) (hexokinase) Level set method (optimisation to solve regularised differential equation) Catherine White & Michael Brady, Proc. Soc. Nuclear Medicine, Toronto, June 2005
Modelling the take up of candidate drugs Segmentation tool Time activity curves of drug concentration Organs of interest segmented Mirada s Research MVS
We have spent a considerable effort developing tools for the effective visualisation of take up of a candidate drug
Mirada s Research MVS software to work with MicroPET We are beginning to image and model the dynamics of drug activity and relate these to cellular and molecular processes
Imaging tumour hypoxia Currently, a range of PET tracers (FMISO, FETNIM, FAZA) are used clinically to image hypoxia; none have optimal characteristics Fundamental problem is that tumour hypoxia is heterogenous and current tracers don t necessarily image this: Need a range of tracers which can image at different levels of hypoxia Alkylating (eg nitromisonidazole): low contrast but repeatable Metal chelates (eg Cu-ATSM) Good contrast, more complex mechanism, harder to make Novel idea: copper/fluorine chemistry bioreductive tracers Alter the reduction potential of a Cu tracer (ATSM) which is labelled with 18 F : cold copper, hot fluorine A range of novel compounds able to image a wide range of hypoxia (Dilworth & Gouverneur in Chemistry) Liposarcoma of left posterior shoulder Left: FDG image Right: F-MISO image
9-4-(18)F-3-(hydroxymethyl)butylguanine Like FLT in that it is a nucleoside analogue, this time for the purine guanine. FHBG enables visualisation of gene expression, a lower level biological process than gene proliferation It is a reporter gene application of PET Processes modeled
Image of FHBG uptake
Simulated Scan NCAT phantom
Generation of a realistic simulation of dynamic PET brain data MNI probabilistic brain atlas FDG model GM, WM time activity curves SORTEO simulator Ecat Exact HR+ Schottlander, Brady, et. al. SPIE 2005 (forthcoming) Fusion of dynamic MR & dynamic PET parametric (PK) maps & cancer typing
An early example: Mirada s Scenium PET atlas, eg from a normal brain(s) coregistered Regions of interest marked up by an expert on the PET atlas Patient PET image Non-rigidly warp the patient image onto the PET atlas Regions of interest deduced on the patient brain
Mirada s Scenium PET atlas, eg from a normal brain(s) coregistered Regions of interest marked up by an expert on the PET atlas Patient PET image Non-rigidly warp the patient image onto the PET atlas Regions of interest deduced on the patient brain Automatically generate a patient specific report, with z- scores and images Compute statistical z- scores of activity in each patient region of interest relative to the atlas level Submit to Web/Grid database manager
Second example: QuantiSPECT QuantiSPECT is an example of a software wrapper for a specific drug Standardised DaTSCAN workflow Quality control Audit trail Clinically validated analysis algorithms Find One Like database
Find One Like allows comparison to normal values. Identify similar scans. Find One Like database can be packaged up for distribution to other sites. Find One Like It The QuantiSPECT approach is generic and can be adapted to other drugs, illnesses, and imaging modalities
Imaging transgene expression noninvasively Luciferase imaging (firefly & Renilla) Can monitor tumour growth/response and metastasis in real time
Image analysis & drug development For every 10,000 lead compounds, one makes it to a product. It costs $1Bn to develop a drug. Can image analysis help? First in human. Preclinical Clinical?? PET & CT & MRI Luminescence imaging MicroPET/CT
Mitosis normal and abnormal cell division From Alberts et al. Molecular Biology of the Cell (2002) Conjecture: abnormal cell divisions (over, say 15 generations) correlates closely with mutations and/or cancerous growth
Fluorescent Microscopy Imaging of Mitosis Images by Rieder et al., Wadsworth Center, NYS Department of Health
Stages of cell division, with a characteristic measurement made at each stage Predicted time course of the measurement over the time period
Automated Image Analysis.... We have a series of images of cell divisions over multiple generations. Our aim is to track each cell (even when they come close to one another, and when they divide) and make suitable measurements. The change in topology leads us to use level sets (a mathematical model of image segmentation) and the uncertainty in the measurements leads us to use an expectationmaximisation approach.
Results Time course of the cell measurement, for a typical set of (normally dividing) cells Ideal model We are now beginning to apply this model to a number of questions in biology, and to cancer development
Colorectal cancer systems biology GE Healthcare + Oxford Improvements in colon cancer imaging Development of colon cancer molecular diagnostics for genomic pathology Genomic signatures in colon cancer Colon cancer molecular diagnostics for chemotherapy selection and prediction of toxicity Information integration All the above in close collaboration with GE Healthcare
Colorectal cancer Original image for multidisciplinary team After removal of bias field Detected nodes 3D reconstruction of colorectum and mesorectal region
Molecular imaging at Oxford A truly rich environment for interdisciplinary research : no silos! World-class faculty, researchers, students Most progressive environment for exploitation of IP in Europe Despite the contributors being in several science departments and in two hospitals, the whole separated by 4Km (uphill) Just imagine what we might achieve if we were all in the same building, adjacent to the cancer hospital and the genetics building