STATEMENT-OF-WORK-EUS cancer detection system 2020 (1).docx

Statement of Work Ultrasound Software Program to Detect Prostate Cancer

CENTER FOR INTERVENTIONAL ONCOLOGY

Statement of Work Ultrasound Software Program to Detect Prostate Cancer 7/18/2020

TABLE OF CONTENTS

1.0 INTRODUCTION/OVERVIEW

The Center for Interventional Oncology seeks to develop novel imaging for many types of solid tumor cancers. The SOW main objective is to develop enhanced US with deep learning for detecting and visualizing tumors during interventional procedures. A contractor is required to do develop a prototype of a spatiotemporal enhanced US (stEUS) based tumor detection using spatial and time (spatiotemporal) methods at 6 months, testing 6-12 months and at 12 months a validated stEUS software program.

1.1 BACKGROUND

Ultrasound (US) imaging due to its real-time imaging property, cost effectiveness, and ease of use is the most common imaging modality for guiding interventional procedures.. However, conventional US imaging offers low sensitivity and specificity in detecting cancer. Transrectal ultrasound imaging of the prostate is 50% accurate in cancer detection, while perhaps more precise imaging e.g., magnetic resonance imaging, is available it is far more costly and for some patients who are claustrophobic or who have an implanted device (e.g., pacemaker) is prohibited. Developing an enhanced transrectal, ultrasound guidance platform that improves both detection and definition of prostate cancer is of great public health interest.

1.2 SCOPE

A contract for computer engineering vendor (academic or commercial) that is skilled in deep learning and medical image computing is required to further developing deep learning methods for stEUS

1.3 OBJECTIVES

Develop a prototype a spatiotemporal enhanced US (stEUS) based tumor detection using spatial and time (spatiotemporal) methods at 6 months, testing 6-12 months and at 12 months a validated stEUS software program.

2.0 REQUIREMENTS

2.1 TASKS

Fiscal Year 2020 work consists of developing a spatiotemporal enhanced US (stEUS) based tumor detection using spatial and time (spatiotemporal) methods. Data from multiple imaging planes and over multiple time-frames of a given region of the prostate will aid in tissue classification (cancer vs. non-cancer). A contract for a computer engineering firm, that is skilled in deep learning and medical image computing is required to further developing deep learning methods for stEUS based tumor detection using primary anonymized images under NIH Clinical Center protocol 10-C- Software expectations are:

a. Accuracy: Area under the ROC curve > 0.75

b. Running time: < 60 second

c. Programming language: Python

d. Platform: Windows, Linux, Mac

2.2 DELIVERABLES/SCHEDULE:

Spatiotemporal enhanced US (stEUS) software for EUS cancer detection system

10 Month Goal Goal 12 month Goal

2.4 ASSUMPTIONS

• Schedule: This 12-month project includes a 6-month milestone to have a prototype software program to initiate preclinical testing. Teleconference meetings with chosen vendor and NIH occurs monthly or more often as deemed appropriate by NIH. At the 12-month mark, a fully validated software package is expected. This software will meet pre-specified requirements as above.

• Resources:

Staff resource: Ph.D. Computer Scientist Programmer, Ph.D. staff support paid for by vendor Skills: expertise is in the areas of ultrasound image analysis and image-guided interventions using machine learning and computer vision techniques The vendor must have access to all the equipment necessary to fulfill the task and deliverables.

Estimated tools and facilities to achieve this include:

Esaote Ultrasound Scanner: Tissue Enhanced Imaging, to improve the quality of ultrasound images by reducing the presence of artifacts, shadowing and speckle. This feature makes it possible for imaging optimization and image analysis in the off-line stage. The scanner supports 2D/3D/4D imaging and is compatible with DICOM and other standards for network data transfer.

Micro-CT Core: resolves dense tissues including bone and contrast-enhanced structures such as vasculature Micro-MRI Core: to perform in vivo studies of animals and ex-vivo studies of excised tissues and samples.

Preclinical Facilities: a state-of-the–art animal housing and research facility designed include nine barriers, quarantine holding rooms and procedure rooms for ABSL-1 and 2 work.

· Proposed Budget: Total: $75,000

Travel expenses to NIH: $6,000

3.0 GOVERNMENT FURNISHED PROPERTY

There is no government furnished property in this SOW.

4.0 SECURITY REQUIREMENTS

The contactor staff must follow all security rules and regulations to gain access to the NIH campus including but not limited to:

1) Required to show one (1) form of identification (a government-issued photo ID-driver's license, passport, green card, etc.) and to state the purpose of their visit

2) Federal law prohibits the following items on Federal property: firearms, explosives, archery equipment, dangerous weapons, knives with blades over 2½ inches, alcoholic beverages and open containers of alcohol.

3) Vendor owned or rented vehicle must go through the Gateway Vehicle Inspection Station (Building 66A), and the vendor must obtain a visitor badge

4) Vendor must contact a Wood Lab staff member for access to the IRL.

image2.png image3.png image1.png