Machine Learning Tool

please resubmit picture if API error message is generated

Disclaimer

This is a 1.0 version of the machine learning model and future updates may occur depending upon user interaction rates and feedback, it cannot be guaranteed 100% accuracy. Upon submitting your picture, the AI will generate an email that Ryan Hennessy is also copied on. Ryan will review the email, and in the event that there is disagreement with the model, he will reach out to you at an estimated time of between 10 minutes to 48 hours.

Note

Please note that this machine learning model is designed as an operational tool for recognizing 1 microbe type or condition at a time per picture and is often NOT representative of overall plant conditions. For any microbe type or condition and its associated growth cause (s) to be relevant it must be ranked common or greater in abundance using the Jenkins ranking system. This model is intended for helping professionals with identification of unknown items they view under the scope. For an overall "big picture" diagnosis of treatment plant conditions a full microscopic evaluation by a trained professional is recommended in which items such as floc structure, dispersed growth, higher life form organisms, filament impact on floc structure, polysaccharide, rank/type of filamentous bacteria types and other indicator organism phenotypes, Gram staining, Neisser staining, and (preferably fluorescent viability testing). Information for sending a sample for microscopic evaluation is available at: Services - Ryan Hennessy Wastewater Microbiology

About

Version 1.0 of this model has been trained using thousands of pictures taken by Ryan Hennessy Wastewater MIcrobiology from a wide range of municipal and industrial wastewater treatment systems throughout predominantly the United States and Canada. The model has been trained for a high variety of microbe types and conditions which include the following list and recommended magnifications for optimal usage:

Microbe Type or Condition	Recommended Photo Options
Actinomycetes_Mycolata (includes Nocardioforms)	1000x phase contrast oil immersion, Gram stain 1000x
Anaerobic sulfur bacteria	1000x phase contrast oil immersion
Anaerobic filament morphotype B	1000x phase contrast oil immersion
Beggiatoa	1000x phase contrast oil immersion or 400x phase contrast
Bristleworm	100x or 200x phase contrast or brightfield
Crustacean	4x, 100x, or 200x phase contrast or brightfield
Dead filament	1000x phase contrast oil immersion
Dispersed growth high	100x or 200x phase contrast
Dispersed growth moderate	100x or 200x phase contrast
Dispersed growth low	100x or 200x phase contrast
Elevated polysaccharide	100x, 200x, 400x reverse India ink
Ferric iron	100x, 200x, 400x, and 1000x phase contrast
Fibrous material	100x or 200x phase contrast, possibly 100x or 200x brightfield (lower probability of success)
Filament bacteria impact on floc structure high	100x or 200x phase contrast, possibly 100x or 200x brightfield and possibly 400x phase contrast and 400x brightfield
Filamentous bacteria impact on floc structure moderate	100x or 200x phase contrast, possibly 100x or 200x brightfield and possibly 400x phase contrast and 400x brightfield
Filamentous bacteria impact on floc structure low	100x or 200x phase contrast, possibly 100x or 200x brightfield and possibly 400x phase contrast and 400x brightfield
Note	It is highly likely that at lower power magnification (100x, 200x) that the machine learning may recognize multiple conditions and lower % (i.e., dispersed growth and floc structure)
Flagellate	100x phase contrast, 100x brightfield, 200x phase contrast, 200x brightfield, 400x phase contrast, 400x brightfield, 1000x phase contrast, 1000x brightfield
Flexibacter	1000x phase contrast, possibly 1000x brightfield
Floc open/diffuse	100x or 200x phase contrast, possibly 100x or 200x brightfield and possibly 400x phase contrast and 400x brightfield
Floc strong	100x or 200x phase contrast, possibly 100x or 200x brightfield and possibly 400x phase contrast and 400x brightfield
Floc weak	100x or 200x phase contrast, possibly 100x or 200x brightfield and possibly 400x phase contrast and 400x brightfield
Free swimming ciliate	100x, 200x, 400x, and 1000x phase contrast and brightfield
Fungi	100x, 200x, 400x, and 1000x phase contrast and likely brightfield
GAO	1000x Neisser
Gastrotrich	100x, 200x, 400x, and 1000x phase contrast
Grease	100x phase contrast cropped, 200x phase contrast 1000x phase contrast
Haliscomenobacter	1000x phase contrast
Hyphomicrobium	1000x phase contrast
Inert material	200x phase contrast possibly 200x brightfield
Iron sulfide	100x phase contrast cropped, 200x, 400x, and 1000x phase contrast
Irregular growth formations	1000x phase contrast
Microscrilla	1000x phase contrast
Microthrix	Gram stain 1000x, 1000x phase contrast, Neisser 1000x (limited success)
Normal polysaccharide	100x, 200x, 400x reverse India ink
Nostocoida limicola	1000x phase contrast, moderate accuracy at 1000x Gram stain and 1000x Neisser stain
Oil	1000x phase contrast
Thiopedia	1000x phase contrast
Thiothrix	1000x phase contrast
Type 021N	1000x phase contrast, possibly 400x phase contrast and 1000x brightfield
Type 0041/0675	1000x phase contrast
Type 0092	Neisser 1000x
Type 0581_Chloroflexi	Best at Gram stain 1000x, low likelihood of recognition at 1000x phase contrast
Type 0914/0803	1000x phase contrast
Type 0961	1000x phase contrast
Type 1851	1000x phase contrast
Type 1863	1000x phase contrast, 1000x Gram stain
Water Bear	100x, 200x, 400x, 1000x phase contrast and brightfield
Yeast	400x phase contrast, 1000x phase contrast
Zoogloea	100x phase contrast (low success), 200x, 400x, and 1000x phase contrast