SOW_Malaria_2016-17.docx

STATEMENT OF WORK

Project Title: Acquisition and Annotation of Malaria Images and Screening software field-test

Synopsis Researchers at the National Library of Medicine (NLM) are developing on a mobile phone-based software tool to speed up malaria screening through analysis of microscopic images of blood smear slides. Professional services are required for the acquisition and annotation of blood smear images to be used for training algorithms and for development of an automated system to detect malaria in the field. Blood smear images should be thick smears as well as thin smears. Thin smears shall be those determined to have P. vivax parasite, while white blood cells shall be both P. falciparum and P. vivax. Professional services are required for field testing of the screening app, and feedback to NLM researchers for improving the capability of the software app.

In the proposed work, the investigator will work closely with the NLM to acquire and annotate specified types of digital blood smear images, and evaluate smartphone mount solutions for extending suitability of the system for field use. The investigator will acquire blood smears images from malaria-infected and uninfected subjects using the smartphone and adapter, and annotate red blood cells in the images as infected or non-infected, white blood cells, parasites, and other objects as appropriate. The location of the subject data collection and quantities and types of blood smears, cell images, and cell annotations are specified within the statement of work.

Description of services

Digitized images of thick and thin blood smears need to be acquired in order to advance the capability of a smartphone-based system for malaria screening. The images are needed for training the statistical machine learning methods so that the automated system can distinguish between red blood cells that are uninfected and red blood cells that are infected by the Vivax malaria parasites. Images are also needed of thick blood smears annotated with white blood cells, parasites, and other relevant objects on the slide. Obtaining accurate counts of the infected cells is a crucial factor of malaria diagnosis and treatment. To achieve the best results, it is important that the acquired images cover the typical shapes, appearances, and frequencies of parasites encountered in the field. It is also important that these images are captured with the same or similar hardware that is used later for the application of the trained system. Therefore, microscopes and smartphone cameras with similar specifications should be used for image acquisition and system deployment. Furthermore, the acquired images need to be annotated manually by expert cell microscopists, so that the system can be trained with typical phenotypes of infected and uninfected red and white blood cells.

Background Malaria is caused by parasites that are transmitted through the bites of infected mosquitoes. With about 200 million cases worldwide, and about 600,000 deaths per year, malaria is a major burden on global health. The greatest impact of the disease is recorded in subSaharan Africa and South East Asia, where a death is recorded every minute from malaria, and where malaria is a leading cause of childhood neuro-disability. Typical symptoms of malaria include fever, fatigue, headaches, and in severe cases seizures, coma, and death.

While existing drugs make malaria a curable disease, inadequate diagnostics and emerging drug resistance are major barriers to successful mortality reduction. The development of a fast and reliable diagnostic test is therefore one of the most promising ways of fighting malaria, together with better treatment, development of new malaria vaccines, and mosquito control.

The current standard method for malaria diagnosis in the field is light microscopy of blood films. About 170 million blood films are examined every year for malaria, which involves manual counting of parasites, and white blood cells that indicate the severity of infection.

Accurate parasite counts are essential to diagnosing malaria correctly, testing for drug-resistance, measuring drug-effectiveness, and classifying disease severity. However, microscopic diagnostics are not standardized and depend heavily on the experience and skill of the microscopist. It is common for microscopists in low-resource settings to work in isolation, with no rigorous system in place that can ensure the maintenance of their skills and thus diagnostic quality. This leads to incorrect diagnostic decisions in the field. For false negative cases, this means unnecessary use of antibiotics, a second consultation, lost days of work, and in some cases progression into severe malaria. For false positive cases, a misdiagnosis entails unnecessary use of anti-malaria drugs and suffering from their potential side effects, such as nausea, abdominal pain, diarrhea, and sometimes severe complications.

To improve malaria diagnostics, the Lister Hill National Center for Biomedical Communications, an R&D division of the US National Library of Medicine, in collaboration with NIH’s National Institute of Allergy and Infectious Diseases (NIAID) and Mahidol-Oxford University, is developing a fully-automated system that uses mobile phone software for parasite detection and counting in microscopic images of blood films.

Automatic parasite counting has several advantages compared to manual counting:

· It provides a more reliable and standardized interpretation of blood films.

· It allows more patients to be served by reducing the workload of the malaria field workers.

· It reduces diagnostic costs.

To count parasites automatically, the system uses image processing methods to find cells infected with parasites in digitized images of blood films. The system first learns the typical shape and visual appearance of infected cells and parasites based on manually-annotated training images. Machine learning methods then discriminate between infected and uninfected cells, detect the type of parasites that are present, and perform the counting. The system uses digital images acquired on standard light microscopy equipment, which makes it well-suited for resource-poor settings. The goal is to use inexpensive and highly portable smartphone technology to acquire blood film images in the field and to run the automated diagnostic system on these images.

Proposed Work

The acquisition of blood smear images should be carried out in a malaria-endemic region. From previous experience in acquiring P. falciparum thin blood smear images, the National Library of Medicine as determined that locations in South and Southeast Asia are endemic with the disease and are suitable sites suitable for acquiring such images. The data acquisition and annotation shall be supervised by a world-class malaria expert that resides and conducts research in the region and works with the field workers for malaria treatment and research. The expert must have existing relationships with the appropriate healthcare entities and research centers in these regions to be considered for technical evaluation. The expert must also have existing relationships with the local government health department to enable acquisition and annotation of blood smear images. This expert will act as the main investigator. The investigator shall have demonstrable prior experience in acquiring and annotating digital images with malaria for thin blood smears. To capture blood smear images, the investigator shall use the built-in camera of a smartphone attached to a conventional microscope by means of an adapter. All essential devices (microscope, smartphone, and adapter) shall be chosen carefully after extensive tests, in tight collaboration between the investigator and NLM researchers. This will ensure that image quality meets the requirements of automated image processing. All devices shall be chosen with later field use in mind, meaning that inexpensive equipment will be preferred when it can provide similar image quality. However, image quality should not be compromised in favor of acquisition costs.

The investigator shall collect 200 peripheral blood slides, among them, 150 infected with malaria with P. vivax and 50 normal with no signs of infection. Each slide will contain a thick and a thin blood smear, so that image processing algorithms can be developed for each smear type, which both play a role in malaria diagnosis. For each thin blood smear, the smartphone camera shall be used to take images and annotate 1000 distinct red blood cells per slide. The thick blood smears on the slides will also be acquired and suitably annotated with 200 white blood cells, parasites, and other objects on every slide.

All images will be manually annotated by microscopists selected by the investigator, who are expert slide readers. They will use an existing online annotation system identified by researchers at NLM. The annotation process will mark the cells by dropping markers in different colors on them. The color will indicate whether a cell is infected or uninfected (e.g. red for infected and green for uninfected cells). A similar procedure shall be followed for thick blood smear images where the microscopists will annotate the white blood cells in the image, parasites, and other objects.

REQUIRED TASKS

Task 1: Adapter and smartphone app evaluation

The investigator shall evaluate various solutions to mount the smartphone for acquiring images of blood slides. These solutions may be recommended by NLM researchers, or could be identified by the investigator. NLM shall provide samples of solutions it identifies. The images shall be acquired through the eyepiece of a microscope, using the camera of the smartphone that is attached to the microscope via the adapter or mount. The goal is to achieve the best image quality possible for manual and automatic cell counting and maximum ease for future reliable use of the setup for field tests and future deployment. The final setup used for image acquisition shall be chosen after consultation with NLM. The investigator may purchase selected smartphone, computer and storage, and other hardware that may be needed for the image acquisition.

The camera settings shall be identical for each image and shall be saved in an image format that shall be agreed on with NLM. The adapter and smartphone shall remain firmly attached to the microscope during the entire image acquisition process. The investigator shall ensure that the microscope used is suitable for imaging blood smears and provides the same image quality as the microscopes in the field.

Task 2: Acquisition and annotation of thick peripheral blood smears from 150 people with P. falciparum type malaria

The investigator shall acquire 150 blood slides with confirmed falciparum malaria infection. The images shall be acquired using the smartphone app provided by the NLM at sites in South or Southeast Asia where this disease is endemic and where routine treatment for such infected people is provided. Each slide of an infected blood smear shall contain a thick film and a thin film. The investigator shall be responsible for the careful preparation of all slides so that automatic image processing is not unnecessarily complicated by incorrect handling of the slides, such as improper staining. The investigator shall ensure that handling of the slides is as close as possible to routine procedures and protocols used in the field. This will ensure that the automatic cell counting tool, which will be trained on these slides, will be readily deployable in the field.

Different colors shall be used to highlight infected and uninfected cells visually. The investigator shall follow the same guidelines as malaria workers use in the field to discriminate and count parasites and white blood cells. A proper and practically relevant manual discrimination is important because it will serve as the training data for the automatic cell counting tool. If needed, the investigator can indicate borderline cases and explain the reason for their classification into the infected or uninfected category. The investigator may also add other useful information, if this will be of future benefit to the project in the investigator’s expert opinion.

For this task only the thick smear images from falciparum infected malaria slides shall be annotated. Thick smear images will at least show white blood cells, and parasites. Parasites, other objects, and 200 white blood cells per slide shall be annotated. The annotation tool shall be identified by the NLM.

Task 3: Acquisition and annotation of thin peripheral blood smears from 150 cases with P. vivax type malaria

For Task 3, the investigator shall acquire and annotate 150 blood infected slides with P. vivax type malaria parasite in addition to the normal uninfected slides collected for Task 1 above.

Different colors shall be used to highlight infected and uninfected cells visually. The investigator shall follow the same guidelines as malaria workers use in the field to discriminate and count infected red blood cells. A proper and practically relevant manual discrimination is important because it will serve as the training data for the automatic cell counting tool. If needed, the investigator can indicate borderline cases and explain the reason for their classification into the infected or uninfected category. The investigator may also add other useful information, if this will be of future benefit to the project in the investigator’s expert opinion.

For this task only the thin smear images shall be acquired and annotated. The images will be acquired using the smartphone app provided by the NLM. The thin smears images show the red blood cells. Infected cells harbor the parasite. 1000 such cells per slide shall be annotated by the investigator. The annotation tool shall be identified by the NLM.

Task 4: Acquisition and annotation of thick peripheral blood smears from 50 people without malaria

The investigator shall acquire 50 confirmed malaria negative blood slides. The images shall be acquired using the smartphone app provided by the NLM at sites in South or Southeast Asia where this disease is endemic and where routine treatment for such infected people is provided. Each slide of an infected blood smear shall contain a thick film and a thin film. The investigator shall be responsible for the careful preparation of all slides so that automatic image processing is not unnecessarily complicated by incorrect handling of the slides, such as improper staining. The investigator shall ensure that handling of the slides is as close as possible to routine procedures and protocols used in the field. This will ensure that the automatic cell counting tool, which will be trained on these slides, will be readily deployable in the field.

Different colors shall be used to highlight the cells, parasites, or other objects visually. In this case, since the images are malaria negative, we do not expect to have any parasites annotated. The investigator shall follow the same guidelines as malaria workers use in the field to discriminate and count parasites and white blood cells. A proper and practically relevant manual discrimination is important because it will serve as the training data for the automatic cell counting tool. If needed, the investigator can indicate borderline cases and explain the reason for their classification into the infected or uninfected category. The investigator may also add other useful information, if this will be of future benefit to the project in the investigator’s expert opinion.

For this task only the thick smear images from malaria negative slides shall be annotated. Thick smear images will at least show white blood cells. Other objects, and 200 white blood cells per slide shall be annotated. The annotation tool shall be identified by the NLM.

Task 5: Acquisition and annotation of thick peripheral blood smears from 150 people with P. vivax type malaria

The investigator shall acquire 150 blood slides with confirmed vivax malaria infection. The images shall be acquired using the smartphone app provided by the NLM at sites in South or Southeast Asia where this disease is endemic and where routine treatment for such infected people is provided. Each slide of an infected blood smear shall contain a thick film and a thin film. The investigator shall be responsible for the careful preparation of all slides so that automatic image processing is not unnecessarily complicated by incorrect handling of the slides, such as improper staining. The investigator shall ensure that handling of the slides is as close as possible to routine procedures and protocols used in the field. This will ensure that the automatic cell counting tool, which will be trained on these slides, will be readily deployable in the field.

Different colors shall be used to highlight infected and uninfected cells visually. The investigator shall follow the same guidelines as malaria workers use in the field to discriminate and count parasites and white blood cells. A proper and practically relevant manual discrimination is important because it will serve as the training data for the automatic cell counting tool. If needed, the investigator can indicate borderline cases and explain the reason for their classification into the infected or uninfected category. The investigator may also add other useful information, if this will be of future benefit to the project in the investigator’s expert opinion.

For this task only the thick smear images from vivax infected malaria slides shall be annotated. Thick smear images will at least show white blood cells, and parasites. Parasites, other objects, and 200 white blood cells per slide shall be annotated. The annotation tool shall be identified by the NLM.

Task 6: Evaluation and Field Testing of the Malaria Screener Software App

The investigator shall test the screener software app for P. falciparum and P. vivax infected images and normal images as it is made available by the NLM. The NLM researchers shall use the images acquired and annotated in Tasks 2 and 3 above to develop cell segmentation and classification software and deploy it in a software app. The smartphone app shall be used by the investigator for field testing. The investigator shall report the performance of the app in terms of its true and false positives and true and false negatives in its ability to screen the disease. The investigator shall provide feedback to NLM which will be used to improve the software capability. After NLM researchers have revised the software, the investigator shall conduct a formal field study with a number of normal and infected patients. The number of patients shall be determined in negotiation with NLM researchers. Results from the field study shall be reported to NLM researchers in terms of the quality of the software using the measures described earlier, and also in terms of the feasibility of using the software (ease of use, reliability, and other use characteristics). The users of the software shall be expert microscopists.

Deliverables

Partial payments will be authorized to the contractor upon receipt of deliverables according to the tasks specified.

The contents of the deliverables shall be as follows:

Deliverable 1 (Year 1): Task 1: Evaluate and select suitable smartphone mounts and adapters for reliable and accurate acquisition of images, and ones that demonstrate potential for future field use.

Deliverable 2 (Year 1): Task 2: De-identified and annotated images of thick smears for 150 slides of P. falciparum malaria infected cases. Parasites, other objects, and 200 white blood cells shall be individually annotated per slide.

Deliverable 1 (Year 2): Task 3: De-identified and annotated images of thin smears for 150 slides infected cases with P. vivax malaria parasite, and de-identified and annotated thin smear images of 50 slides of normal cases. The images must have at least 1000 cells annotations per slide.

Deliverable 2 (Year 2): Task 4: De-identified and annotated images of thick smears for 50 malaria negative cases. The images must have 200 cell annotations per slide.

Deliverable 3 (Year 2): Pilot evaluation of existing prototype of NLM MalariaScreener app and recommend user interface improvements and early performance issues. The investigator shall travel to NLM for design discussions and provide recommendations to NLM researchers. The NLM researchers shall take the opportunity to observe the investigator using the app on a microscope and take notes to improve usability for future field use

Deliverable 1 (Year 3): Task 5: De-identified and annotated images of thick smears for 150 slides of P. vivax malaria infected cases. Parasites, other objects, and 200 white blood cells shall be individually annotated per slide.

Deliverable 2 (Year 3): Task 6: Design and conduct a formal evaluation of the NLM MalariaScreener app in the field and provide a performance report on its use by a number of field workers. The performance report shall include the technical performance of the smartphone app, and also include other information that can assist in advancing the capabilities of the app. Details of the report format can be negotiated with the Government at a suitable time.

Each deliverable above shall include a comprehensive report on all work, in summary form and in detail, theoretical and heuristic rationale for the approach taken, conclusions, recommendations and references.

The Contractor shall apprise the Government of costs involved for any hardware and related supplies needed to commence the work. This amount may be invoiced immediately upon being awarded the contract. The remainder amount of the award for the deliverables listed in this statement of work shall be made at 6 months or on completion of the deliverable, whichever comes sooner.

Notice of Government Unlimited Rights to Work First Produced Under This Contract

Government rights to work first produced under this contract are established by Federal law including, but not limited to, this specific reference: FAR 52.227-14, – General, (b) (1).

Requirement to Notify Government of Proprietary Work Dependencies

The Contractor is required to notify the Government in writing of any dependencies of the deliverables under this contract on proprietary, copyrighted, or patented work that potentially inhibits, restricts, or requires permission for the dissemination of the deliverables to the public, other governmental agencies or research groups, or to any other parties whatsoever.

Justification

Professional services of experts from an entity involved in malaria research in endemic regions of south-east Asia, in particular Bangladesh and Thailand, are required to support ongoing research into designing an automated cell counting tool for malaria diagnosis and clinical decision support. This research activity is an ongoing project at the National Library of Medicine following its inception from initial support provided by US Department of Health and Human Services through its HHS Innovation Ventures Fund Program.

Evaluation Criteria

1. The Contractor shall be a known expert in the areas of malaria diagnosis and pathogenesis (by peer-reviewed conference and scientific journal publications), and shall have demonstrable prior experience in acquiring and annotating blood smear images using smartphone affixed on an standard optical microscope in the malaria endemic regions identified in this SOW. The investigator shall be an expert in combining clinical studies, descriptive epidemiology and mathematical modelling of malaria in South and Southeast Asia. The investigator’s expertise must include population dynamic mathematical modelling of artemisinin resistance and malaria elimination; spatiotemporal epidemiology, GIS mapping and population movement; as well have ongoing clinical studies on artemisinin resistance, observational studies on pathogenesis, malaria retinopathy and medical imaging in severe malaria; clinical trials of treatment and supportive care for severe malaria and diagnosis of non-malaria causes of fever. Demonstrated collaborations with multi-national partners will be considered an advantage due to the global health threat malaria poses. Ongoing collaboration with the Malaria Research Unit at the National Institute of Allergic and Infectious Diseases (NIAID) at the National Institutes of Health (NIH) is an added bonus.

2. The investigator shall have ongoing research efforts and tight collaborations with expert microscopists and workers in the field, and have existing relationships with healthcare and research authorities in South and Southeast Asia, in particular in Bangladesh and Thailand, and have the necessary authorization to acquire images of patient blood smears in these regions. In addition, the investigator needs to have the capability and technical facilities to collect a large number of blood slides over the period of performance.

3. Prior experience in using the online annotation for annotating P. Falciparum infected malaria thin blood smears is required.

Period of performance

The period of performance shall be for 12 months from date of award with two (2) successive 12-month Option Years.

Place of Performance

TBD