Written by Diana Perez Staples, Ph.D., and Horacio Tapia McClung, Ph.D.
The Mediterranean fruit fly is one of the most harmful pests in the world (Ceratitis capitataMexican fruit flyAnastriva Ludens), causing billions of dollars in damage to agriculture. Fortunately, sterile insect technology is currently being used as part of integrated area management programs to control these flies in certain regions of the world.
The sterile insect technique (SIT) is a form of birth control, and it consists of raising millions of these flies in factories, irradiating them with X-rays or gamma rays to make them sterile, and then releasing them into areas where the pests are. When sterile males mate with wild females, the females will not have fertile eggs to lay in the fruits. Thus, population levels decrease. SIT has good green credentials because it only targets pest species, does not introduce foreign genetic material into the population, and it reduces the use of pesticides.
The irradiation process in the decimal sterilization technique is the key to its success. For teeming flies, irradiation is usually performed 2 days before the pupae emerge as adults. If the pupae are irradiated too early or too late in their development process, this can lead to problems with movement and behavior as adults. However, even during controlled conditions, the pupae can vary in their growth time. Thus, one of the tests that is performed before irradiation is to determine the physiological age of the virgins.
Currently, in these fruit fly plants around the world, technicians must determine the appropriate time to irradiate by taking a sample of the pupae, removing the pupal case to reveal the eyes, and then checking eye color against the color chart. This can be daunting and subject to human error, as it depends on the skill, experience and expertise of the technician, as well as natural biases in interpreting colors. Technicians can get tired of this repetitive work, while sick days and vision problems can also cause differences in correct identification.
For this, as part of a Ph.D. thesis. At the Cencias Agricolas College of the University of Veracruzana, Ivan Gonzalez Lopez, now at the IAEA Entomology Laboratory and the Food and Agriculture Organization in Austria, captured images of the exposed pupae eyes of both Mediterranean fruit flies and Mexican fruit flies. We chose virgins who still had a few days to emerge and purposely took ragged photos that did not have ideal lighting conditions or focus. In fact, they were taken quickly and with a mobile phone.
Then, as part of a master’s research at Laboratorio Nacional de Informática Avanzada in Xalapa Veracruz, Georgina Carrasco processed images with software trained to detect and crop the eye area in the image. Then, using correct answers from a factory technician, another algorithm was trained through a supervised machine learning method known as transfer learning, to accurately determine the age of the pupae.
We found that algorithms based on a neural network architecture known as Inception v1 correctly determined the physiological age of maturity at 2 days before emergence, with an accuracy of 75% for Mexican Drosophila and 83.16% for Mediterranean Drosophila, respectively. This method is certainly not perfect, and is still technically demanding to dissect the cocoons and capture images, but it is a promising approximation of how supervised machine learning and artificial intelligence can be used to aid uncertainty in decisions about irradiation time. The level of resolution can also be improved as more images are taken and provided for the algorithm to learn from.
The next steps will be to develop software that technicians can easily use as well as to train these algorithms with other types of tephritis that are currently controlled through SIT technology. Certainly, it highlights that there could be some exciting collaboration between entomologists and AI researchers.
Diana Perez Staples, Ph.D., Research Professor at the Institute of Biotechnology and Applied Ecology at the University of Veracruzana, Xalapa, Veracruz, Mexico. E-mail: firstname.lastname@example.org. Horacio Tapia McClung, Ph.D., is a research professor at the Artificial Intelligence Research Institute of the University of Veracruzana also in Xalapa. E-mail: email@example.com.