| --- |
| tags: |
| - image-classification |
| - microbiology |
| - gram-stain |
| - medical-imaging |
| - research-use-only |
| pipeline_tag: image-classification |
| --- |
| |
| # BiTTE-lite |
|
|
| <p align="center"> |
| <img src="./icon.jpg" alt="BiTTE-lite icon" width="220"> |
| </p> |
|
|
| BiTTE-lite is an application within the CarbConnect platform. It is a simplified version of the BiTTE product, designed for easy and efficient detection of microorganisms in various applications. |
|
|
| The app focuses on classifying microorganisms from Gram-stained microscopy images into seven output categories: |
|
|
| 1. Gram-negative rods (GNR) |
| 2. Gram-negative cocci (GNC) |
| 3. Gram-positive rods (GPR) |
| 4. Gram-positive cocci (GPC) |
| 5. Yeast-like fungi (Yeast) |
| 6. No bacteria |
| 7. Combination / mixed findings |
|
|
| ## Intended Uses and Limitations |
|
|
| BiTTE-lite is strictly intended for **Research Use Only (RUO)**. |
|
|
| It is **not** intended for: |
|
|
| - clinical diagnostics |
| - medical decision-making |
| - patient management |
| - therapeutic selection |
| - any regulated medical procedure |
|
|
| For additional product details, refer to the BiTTE-lite Learn More page on CarbConnect. |
|
|
| ## How to Use |
|
|
| A video tutorial demonstrating how to use the app is available on YouTube. |
|
|
| ## Training Data |
|
|
| The model was trained on a dataset of Gram-stained images of urine and blood culture specimens provided by: |
|
|
| - the School of Medicine, Kobe University |
| - the National Center for Global Health and Medicine (NCGM) |
|
|
| Specimens were Gram-stained using either the Favor or Barmy method. |
|
|
| Image acquisition was performed by photographing specimens through the eyepiece of an optical microscope at **1000x magnification** using a smartphone camera. |
|
|
| The dataset captures frequently encountered clinical bacterial species and includes: |
|
|
| - 15 species in urine specimens |
| - 19 species in aerobic blood culture specimens |
| - 13 species in anaerobic blood culture specimens |
|
|
| ## Performance and Evidence |
|
|
| Related publication: |
|
|
| Kei Yamamoto, Goh Ohji, et al. *Accuracy of classification of urinary Gram-stain findings by a computer-aided diagnosis app compared with microbiology specialists*. J Med Microbiol. 2025 Apr;74(4):002008. doi: 10.1099/jmm.0.002008. |
|
|
| Paper link: |
|
|
| https://www.microbiologyresearch.org/content/journal/jmm/10.1099/jmm.0.002008 |
|
|
| ## Citation |
|
|
| If you use BiTTE-lite in research, please cite: |
|
|
| ```bibtex |
| @article{yamamoto2025bitte, |
| author = {Yamamoto, Kei and Ohji, Goh and others}, |
| title = {Accuracy of classification of urinary Gram-stain findings by a computer-aided diagnosis app compared with microbiology specialists}, |
| journal = {Journal of Medical Microbiology}, |
| year = {2025}, |
| month = {Apr}, |
| volume = {74}, |
| number = {4}, |
| pages = {002008}, |
| doi = {10.1099/jmm.0.002008} |
| } |
| ``` |
|
|
| ## Other Remarks |
|
|
| For detailed classification at the species level, refer to **BiTTE - iE**. |
|
|
| For guidance on achieving high-quality Gram staining, refer to the automated gram stainer **Point of Care Gram Stainer (PoCGS)**. |
|
|
