Behavioural analysis and modification underpin APOPO's training process

APOPO is carrying out applied behavioural research towards optimizing both the training process and the performance of detection rats. These are affected by numerous parameters relating to aspects of learning, olfaction, odour stimulus, reinforcement schedules, and environmental influences amongst others.

Key research areas include the reduction of training times and the standardisation of the training process, starting from socialization of the juvenile rat through final accreditation as part of the detection system:

• Acquisition of the desired stimulus control on the target scent has also been one of the main challenges throughout the development of the detection systems. This has involved training through a large variety of background odours, while controlling for unwanted cues.
• Research has also been conducted into the sensitivity of the detection rats on various stimuli of interest. This remains an enormous challenge as their sensitivity is often beyond that of the analytical equipment used for confirmation. Extreme control is required over the presentation and composition of the target scent, which is often not a single substance but rather a "bouquet".
• In order to enhance learning speed, while maintaining the desired search behaviour, during operations where the search targets are unknown to the trainer, appropriate reinforcement schedules have to be implemented at each training stage.

For the Mine Detection Rats, variations were observed depending on mine type, search pattern, search speed, vegetation and climate. These influence the decisions to be made on when, how, and under which conditions the animals should be trained and used in operations.

APOPO innovates the olfactometer for rapid sample screening

APOPO is also researching a variety of scent detection configurations and olfactometers for optimizing sample evaluation by the rats.

Initial efforts focused on developing a fully automated explosive vapour delivery system attached to a Skinner box, with a typical two-lever press protocol. This was in co-operation with the Canadian lab "Jenel TVD". However, limitations in the variability and quantification of automated scent presentation through a central sniffing hole led to APOPO investigating alternative setups.

This resulted in the initial development of the Line Cage, which is still in use for REST and TB sample evaluation. The cage has a stainless steel floor with a linear array of sniffing holes, through which the scent samples are exposed. The samples are presented in a cassette, containing all samples for one run, thus avoiding individual handling of the scent samples. Upon correct indication on a sample, the rat receives a "click" followed by a food reward delivered at one of the cage ends, before resuming its search from the point of indication. Using this system a detection rat can evaluate approximately 100 scent samples in 15 minutes.

An alternative design consists of a wall-attached, stainless steel tunnel punctuated with number of sniffing holes. The samples are individually placed under each hole and un-covered as the rat approaches. This system needs two sample changers continuously inserting and opening samples ahead of the rat as it progresses. Upon the correct indication of a sample, the trainer delivers the food reinforcer just after the location of the indicated sample.

A recent development is an automated, circular setup that is designed to provide distinct individual sample presentations without a spatial bias. The increased separation of the samples ensures that no cross-contamination of scent can occur between them. The sniffing holes include optical sensors that register a rat's indication, through the breaking of a light-beam for a preset duration. This automatically triggers an electronic clicker and the delivery of a food reinforcer.

Further development of the line cage is underway to automatically register the rats' response and deliver the food reinforcement, in an effort to remove the human factor and improve the standardization of the training protocol.

Frequent analyses of the functional characteristics of each of the designs is essential, as this aids in the further improvement and standardization of APOPO's evaluation and training setup.

The race to trace detection: rats in tandem with analytical chemistry

Landmine Detection

Analytical Environmental Chemistry plays a pivotal role in studying the environmental factors which affect the transport of explosive compounds through the soil/water and soil/air interfaces. Factors such as soil type, soil moisture content, soil temperature, duration and intensity of precipitation, vegetation, type of mine, etc. all influence the amount of target scent available for animal detection. As trained rats detect scent emanating from buried landmines, knowing the quality and quantity of volatile explosive compounds around the suspected area is of paramount importance to assist in animal training and testing.

The chemistry team conducts research on soil samples collected from above landmines buried in APOPO's test field, while recording the corresponding environmental parameters from a weather station with sensors installed around the field. The collected samples undergo extraction and analytical testing for explosive compounds, and the quality and quantity of these compounds are then correlated with their respective environmental parameters. The aim is to deduce a simple algorithm that may assist in sample collection and animal detection.

The chemistry team also contributes to the design of Remote Sampling Devices and equipment, through analyzing various materials for use as filters. This helps in selecting the optimal material for sampling above buried landmines, so that explosive scents are selectively, and effectively, absorbed onto the material for remote animal and instrumental detection.

The chemistry team has also been investigating the volatile compounds in Tuberculosis (TB) patient sputum samples, which trigger a response from the detection rats. These volatiles are extracted from both confirmed positive sputum samples, as well as TB culture colonies, using Solid Phase Micro Extraction (SPME). The extracts are then analyzed using Gas Chromatography-Mass Spectroscopy (GCMS). This research helps to fine-tune the animal training as well as improve detection.

In this ongoing area of study, negative samples are spiked with Tuberculosis and other micro-bacteria to investigate whether the detection rats can discriminate between pathogenic and non-pathogenic strains. Detection rats are also tested on pure volatile compounds, previously identified as TB markers, such as culture samples.

Research is also being conducted to study the scientific, technical, and clinical feasibility of breath-sampling especially for mass screening in congested environments like slums, prisons and schools.

Remote scent collection: bringing the smell to the rat.

APOPO is conducting a series of studies on sampling equipment used for REST operations and training. These studies are at the behest of the GICHD, who provide the main source of funding.

The studies comprise:

• Analyses of the characteristics and variability of the coiled PVC wire filters developed and used by MECHEM, and also used in "RASCO" for airport security purposes;
• Identification, development, and testing of filter materials to be considered for REST sampling;
• Identification, development, and testing of sampling pumps;
• Identification, development, and testing of sample containers.

The initial studies resulted in the improvement of filter materials, packing materials, and backpack pumps. The studies also showed that it was more reliable to collect surface dust, rather than depend on explosives in the vapour phase, which tend to be very unstable due to wind and convection.

APOPO developed two surface dust collection devices in collaboration with the Belgian Design and Study Bureau, Etoile Mechanique. These devices were designed to collect only dust, without the use of filter materials.

Two different prototypes were made:

• One prototype uses compressed gas delivered through a jet flow nozzle to stir up the dust from among the vegetation, then the dust is collected;
• The other prototype uses simple front brushes to stir up the dust, and consequently collects the airborne dust particles using a vortex particle separator.

This latter prototype is currently in use by APOPO for collecting the limited amounts of samples used for research purposes.