Correlation of pre-operative cancer imaging techniques with post-operative macro and microscopic lung pathology images

Gabriel Reines March, Xiangyang Ju, Stephen Marshall, Stephen Harrow, Craig Dick

Research output: Contribution to conferencePoster

Abstract

Background:
This research project aims to investigate the performance of several PET radiotracers in lung cancer by aligning PET-CT and pathology imagery acquired from the same patients at different points in time. The discrimination of tumour substructures is of great importance in therapy planning, as a given treatment may be better adapted depending on the local characteristics of the carcinoma.
Design:
Due to the high deformability of lung tissue, several intermediate steps must be used for merging pathology and pre-operative PET-CT in a coherent manner. Firstly, the tumour volume is reconstructed from the macroscopic images taken during dissection. For this purpose, an enhanced dissection protocol is used, where the lung specimen is placed in a bespoke slicing rig and embedded in agar to hold it in place. Using a threaded plunger, the specimen is pushed upwards in 5mm steps, sliced and photographed. This procedure allows us to obtain slices of uniform thickness. Secondly, microscopic digital slides of the cancerous tissue are merged with the macroscopic 3D model. Next, the whole volume is aligned to an ex-vivo CT scan of the specimen, taken immediately before pathological dissection. Finally, the pre-registered volume is fused with the pre-operative PET-CT scan, using a non-linear deformable model.
Results:
Preliminary results obtained with a synthetic phantom allowed us to analyse the accuracy of the tumour 3D reconstruction algorithm from planar macroscopic slices. Using these findings, we could optimise the interpolation and segmentation routines for building an accurate 3D model of the carcinoma. During our first trial with lung tissue (on-going work), each cross-sectional slice was photographed, the tumour boundary was delineated in each image by a pathologist (CD), and from these contours a high-resolution 3D tumour model was built. Next, the corresponding microscopic digitised slices were merged. To date, five further patients have been identified and consented, therefore allowing us to test our algorithm on different cases and assess its performance.
Conclusion:
We demonstrate a novel set of methods for co-registration of pre-operative PET-CT to macro and microscopically defined lung tumours. This proof of principle now allows interrogation of the raw data from scans using a range of tracers and the development of algorithms that identify substructure detail within a tumour mass, which could lead to tailored radiotherapy for individual cases based on tracer patterns and uptake.

Conference

ConferenceUSCAP 2018 Annual Meeting
Abbreviated title107th Annual Meeting
CountryCanada
CityVancouver
Period17/03/1823/03/18
Internet address

Fingerprint

Post and Core Technique
Pathology
Lung
Dissection
Neoplasms
Carcinoma
Nonlinear Dynamics
Imagery (Psychotherapy)
Tumor Burden
Agar
Lung Neoplasms
Radiotherapy
Positron Emission Tomography Computed Tomography
Therapeutics
Research

Keywords

  • cancer
  • imaging
  • pathology

Cite this

Reines March, G., Ju, X., Marshall, S., Harrow, S., & Dick, C. (Accepted/In press). Correlation of pre-operative cancer imaging techniques with post-operative macro and microscopic lung pathology images. Poster session presented at USCAP 2018 Annual Meeting, Vancouver, Canada.
Reines March, Gabriel ; Ju, Xiangyang ; Marshall, Stephen ; Harrow, Stephen ; Dick, Craig. / Correlation of pre-operative cancer imaging techniques with post-operative macro and microscopic lung pathology images. Poster session presented at USCAP 2018 Annual Meeting, Vancouver, Canada.
@conference{dbcc0870552545b18751fd598da0575e,
title = "Correlation of pre-operative cancer imaging techniques with post-operative macro and microscopic lung pathology images",
abstract = "Background:This research project aims to investigate the performance of several PET radiotracers in lung cancer by aligning PET-CT and pathology imagery acquired from the same patients at different points in time. The discrimination of tumour substructures is of great importance in therapy planning, as a given treatment may be better adapted depending on the local characteristics of the carcinoma.Design:Due to the high deformability of lung tissue, several intermediate steps must be used for merging pathology and pre-operative PET-CT in a coherent manner. Firstly, the tumour volume is reconstructed from the macroscopic images taken during dissection. For this purpose, an enhanced dissection protocol is used, where the lung specimen is placed in a bespoke slicing rig and embedded in agar to hold it in place. Using a threaded plunger, the specimen is pushed upwards in 5mm steps, sliced and photographed. This procedure allows us to obtain slices of uniform thickness. Secondly, microscopic digital slides of the cancerous tissue are merged with the macroscopic 3D model. Next, the whole volume is aligned to an ex-vivo CT scan of the specimen, taken immediately before pathological dissection. Finally, the pre-registered volume is fused with the pre-operative PET-CT scan, using a non-linear deformable model.Results:Preliminary results obtained with a synthetic phantom allowed us to analyse the accuracy of the tumour 3D reconstruction algorithm from planar macroscopic slices. Using these findings, we could optimise the interpolation and segmentation routines for building an accurate 3D model of the carcinoma. During our first trial with lung tissue (on-going work), each cross-sectional slice was photographed, the tumour boundary was delineated in each image by a pathologist (CD), and from these contours a high-resolution 3D tumour model was built. Next, the corresponding microscopic digitised slices were merged. To date, five further patients have been identified and consented, therefore allowing us to test our algorithm on different cases and assess its performance.Conclusion:We demonstrate a novel set of methods for co-registration of pre-operative PET-CT to macro and microscopically defined lung tumours. This proof of principle now allows interrogation of the raw data from scans using a range of tracers and the development of algorithms that identify substructure detail within a tumour mass, which could lead to tailored radiotherapy for individual cases based on tracer patterns and uptake.",
keywords = "cancer, imaging, pathology",
author = "{Reines March}, Gabriel and Xiangyang Ju and Stephen Marshall and Stephen Harrow and Craig Dick",
year = "2017",
month = "10",
day = "30",
language = "English",
note = "USCAP 2018 Annual Meeting : Geared to Learn, 107th Annual Meeting ; Conference date: 17-03-2018 Through 23-03-2018",
url = "https://www.uscap.org/meetings/detail/2018-annual-meeting/",

}

Reines March, G, Ju, X, Marshall, S, Harrow, S & Dick, C 2017, 'Correlation of pre-operative cancer imaging techniques with post-operative macro and microscopic lung pathology images' USCAP 2018 Annual Meeting, Vancouver, Canada, 17/03/18 - 23/03/18, .

Correlation of pre-operative cancer imaging techniques with post-operative macro and microscopic lung pathology images. / Reines March, Gabriel; Ju, Xiangyang; Marshall, Stephen; Harrow, Stephen; Dick, Craig.

2017. Poster session presented at USCAP 2018 Annual Meeting, Vancouver, Canada.

Research output: Contribution to conferencePoster

TY - CONF

T1 - Correlation of pre-operative cancer imaging techniques with post-operative macro and microscopic lung pathology images

AU - Reines March, Gabriel

AU - Ju, Xiangyang

AU - Marshall, Stephen

AU - Harrow, Stephen

AU - Dick, Craig

PY - 2017/10/30

Y1 - 2017/10/30

N2 - Background:This research project aims to investigate the performance of several PET radiotracers in lung cancer by aligning PET-CT and pathology imagery acquired from the same patients at different points in time. The discrimination of tumour substructures is of great importance in therapy planning, as a given treatment may be better adapted depending on the local characteristics of the carcinoma.Design:Due to the high deformability of lung tissue, several intermediate steps must be used for merging pathology and pre-operative PET-CT in a coherent manner. Firstly, the tumour volume is reconstructed from the macroscopic images taken during dissection. For this purpose, an enhanced dissection protocol is used, where the lung specimen is placed in a bespoke slicing rig and embedded in agar to hold it in place. Using a threaded plunger, the specimen is pushed upwards in 5mm steps, sliced and photographed. This procedure allows us to obtain slices of uniform thickness. Secondly, microscopic digital slides of the cancerous tissue are merged with the macroscopic 3D model. Next, the whole volume is aligned to an ex-vivo CT scan of the specimen, taken immediately before pathological dissection. Finally, the pre-registered volume is fused with the pre-operative PET-CT scan, using a non-linear deformable model.Results:Preliminary results obtained with a synthetic phantom allowed us to analyse the accuracy of the tumour 3D reconstruction algorithm from planar macroscopic slices. Using these findings, we could optimise the interpolation and segmentation routines for building an accurate 3D model of the carcinoma. During our first trial with lung tissue (on-going work), each cross-sectional slice was photographed, the tumour boundary was delineated in each image by a pathologist (CD), and from these contours a high-resolution 3D tumour model was built. Next, the corresponding microscopic digitised slices were merged. To date, five further patients have been identified and consented, therefore allowing us to test our algorithm on different cases and assess its performance.Conclusion:We demonstrate a novel set of methods for co-registration of pre-operative PET-CT to macro and microscopically defined lung tumours. This proof of principle now allows interrogation of the raw data from scans using a range of tracers and the development of algorithms that identify substructure detail within a tumour mass, which could lead to tailored radiotherapy for individual cases based on tracer patterns and uptake.

AB - Background:This research project aims to investigate the performance of several PET radiotracers in lung cancer by aligning PET-CT and pathology imagery acquired from the same patients at different points in time. The discrimination of tumour substructures is of great importance in therapy planning, as a given treatment may be better adapted depending on the local characteristics of the carcinoma.Design:Due to the high deformability of lung tissue, several intermediate steps must be used for merging pathology and pre-operative PET-CT in a coherent manner. Firstly, the tumour volume is reconstructed from the macroscopic images taken during dissection. For this purpose, an enhanced dissection protocol is used, where the lung specimen is placed in a bespoke slicing rig and embedded in agar to hold it in place. Using a threaded plunger, the specimen is pushed upwards in 5mm steps, sliced and photographed. This procedure allows us to obtain slices of uniform thickness. Secondly, microscopic digital slides of the cancerous tissue are merged with the macroscopic 3D model. Next, the whole volume is aligned to an ex-vivo CT scan of the specimen, taken immediately before pathological dissection. Finally, the pre-registered volume is fused with the pre-operative PET-CT scan, using a non-linear deformable model.Results:Preliminary results obtained with a synthetic phantom allowed us to analyse the accuracy of the tumour 3D reconstruction algorithm from planar macroscopic slices. Using these findings, we could optimise the interpolation and segmentation routines for building an accurate 3D model of the carcinoma. During our first trial with lung tissue (on-going work), each cross-sectional slice was photographed, the tumour boundary was delineated in each image by a pathologist (CD), and from these contours a high-resolution 3D tumour model was built. Next, the corresponding microscopic digitised slices were merged. To date, five further patients have been identified and consented, therefore allowing us to test our algorithm on different cases and assess its performance.Conclusion:We demonstrate a novel set of methods for co-registration of pre-operative PET-CT to macro and microscopically defined lung tumours. This proof of principle now allows interrogation of the raw data from scans using a range of tracers and the development of algorithms that identify substructure detail within a tumour mass, which could lead to tailored radiotherapy for individual cases based on tracer patterns and uptake.

KW - cancer

KW - imaging

KW - pathology

UR - https://www.uscap.org/meetings/detail/2018-annual-meeting/

M3 - Poster

ER -

Reines March G, Ju X, Marshall S, Harrow S, Dick C. Correlation of pre-operative cancer imaging techniques with post-operative macro and microscopic lung pathology images. 2017. Poster session presented at USCAP 2018 Annual Meeting, Vancouver, Canada.