Spoti-find: A novel, open-source void spot assay image analysis tool

Abstract

The void spot assay (VSA) is a widely used, non-invasive method for evaluating urinary behavior in rodents, but existing analysis tools are limited in scope, throughput, or accessibility. We developed Spoti-find, a stand-alone, open-source VSA image analysis application that introduces novel, biologically meaningful metrics including void circularity, distance to paper edge, and volume-based binning into primary, micro-, and nanovoids. Designed with usability in mind, Spoti-find features a graphical interface that enables manual or semi-automated spot identification, adjustable thresholds, and streamlined data export without the need for coding expertise. We validated Spoti-find across diverse datasets, showing strong inter-user consistency, sensitivity to known phenotypes in aging and disease models, and accuracy in capturing novel parameters while demonstrating high agreement with existing tools. By capturing behavioral context and spatial morphology in voiding patterns, Spoti-find expands the interpretive power of VSA and provides a flexible, user-friendly platform for phenotyping urinary dysfunction in preclinical studies.

NEW & NOTEWORTHY

The VSA is a popular, non-invasive method for studying rodent urinary behavior, but current tools lack flexibility and require extensive manual quality control. We developed Spoti-find, an open-source application introducing novel parameters like void circularity, edge proximity, and volume-based binning. With a user-friendly interface requiring no coding, Spoti-find enables manual and semi-automated analysis with high consistency and transparency.

Graphical Abstract

Urinary dysfunction is a hallmark of numerous neurological, metabolic, and age-related diseases, yet the biological basis of aberrant voiding behavior remains difficult to study in preclinical models. The void spot assay (VSA) provides a non-invasive, low-cost method to assess urinary behavior in rodents by capturing the number, size, and distribution of urine spots deposited on filter paper over time (1). Its simplicity, flexibility, and relevance to multiple organ systems -- including the bladder, kidney, central nervous system, and musculoskeletal system -- make it a powerful tool for investigating urinary phenotypes across diverse experimental contexts.

The most widely used platform and current gold standard, Void Whizzard, was a pivotal advancement for the field. By offering semi-automated analysis of VSA images within ImageJ, it made standardized spot detection, sizing, and classification broadly accessible for the first time (2). More recently, ML-UrineQuant extended these capabilities by incorporating deep learning for high-throughput analysis, enabling automated spot classification across large datasets with adaptable parameters tailored to specific lab needs (3). Despite these advances, both platforms have limitations: they are limited in capturing novel biologically relevant metrics with clinical implications, often require extensive troubleshooting and quality control (QC) checks by the experimenter, and in the case of machine learning–based approaches, the opaque ‘black box’ nature of the algorithms can make resolving issues challenging.

To address the need for biologically meaningful VSA analysis that remains accessible and easy to use, we developed Spoti-find, a stand-alone, open-source Python-based application. Spoti-find builds upon prior tools by introducing analytical parameters that capture the spatial and morphological context of voiding behavior, features that may reflect underlying physiological or behavioral states. These include: (i) void circularity, a proposed proxy for intentional voiding versus passive leakage; (ii) distance from the edge of the filter paper, which may reflect anxiety or urgency-like phenotypes (4) based on rodents’ preference for voiding along boundaries (5); and (iii) void size binning, which classifies spots into primary voids, microvoids, and nanovoids using biologically interpretable volume cutoffs. Designed with usability in mind, Spoti-find features an intuitive graphical interface, allows both automated (rectangle) and manual (polygon) spot selection, enables adjustable thresholds, and streamlines data export. While different platforms may better suit different experimental goals—such as ultra-high throughput versus detailed spatial metrics -- Spoti-find offers a customizable and behaviorally informative alternative that expands the analytical utility of VSA for diverse preclinical applications.

Importantly, Spoti-find was designed for accessibility and transparency. The tool requires no coding knowledge, features a graphical interface, and allows adjustable thresholds for all void classification parameters. While Spoti-find is a manual program and does not prioritize high throughput, it enables rigorous data extraction with minimal quality control effort, even across variable experimental conditions. In this study, we validate Spoti-find across multiple independent datasets and demonstrate its ability to detect known urinary phenotypes, while uncovering additional spatial and morphological features that may offer translational insights. By emphasizing interpretability, flexibility, and biological relevance, Spoti-find represents an approach to VSA analysis that bridges the gap between simple descriptive metrics and high-dimensional behavioral phenotyping.

MATERIALS AND METHODS

Animals

For the data collected at The University of Wisconsin, C57BL/6J animals were obtained from The Jackson Laboratory and maintained on corn cob bedding and given a standard rodent diet (Teklad 2018). The room was maintained between 22 and 24 °C with a 12-hour light/dark cycle. For the data collected at The Jackson Laboratory, animals were maintained on aspen shavings and given a standard rodent diet (LabDiet 5KOG) and acidified water in a pathogen free room. The room was maintained at 21°C with a 12-hour light/dark cycle (6am to 6pm). All animal experiments were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (National Research Council) and were approved by The University of Wisconsin and The Jackson Laboratory’s Animal Care and Use Committees.

Void Spot Assay

At the University of Wisconsin, mice were placed individually on 16 x 26-cm thick chromatography paper (Ahlstrom, Kaukauna, WI) secured to standard mouse cages. Mice were restricted from water during the 4-hour study duration. Filter papers were imaged with a BioRad ChemiDoc Imaging System under ultraviolet light using an ethidium bromide filter set and 0.5 second exposure (BioRad, Hercules, CA).

At The Jackson Laboratory, mice were placed in a cage lined with filter paper (Grade 328 from Alstrom 2381-1417) for 4 hours while restricted from water. After the animal was removed, the paper was dried for a minimum of 24 hours and exposed to UV. An image was taken using a Syngene Ingenious bioimaging system (Syngene International).

Program development

Spotifind was developed in Python. The code and supporting documents, including the installation instructions, user manual, and validation images can be found at https://doi.org/10.5281/zenodo.18622284. Briefly we use intensity thresholds to identify the paper from background, if the default settings do not work well with the lighting of experimental papers this can be adjusted in the program. Urine spots are identified either using a threshold-based method with the rectangle tool or manually using the polygon tool. Once a spot is identified several measurements are taken. Each spot is counted and classified based on number of pixels compared to a calibration data sheet. Additionally, eccentricity and distance from the center of the spot to nearest edge are calculated.

The data in this paper was analyzed according to the user manual. A combination of rectangle tool and manual polygon spot identification was used. To ensure that users identify spots with similar accuracy and precision, we created a validation data set that is available at https://github.com/TheJacksonLaboratory/Spoti-find and also includes a file with results from three different observers and R code to calculate the interclass correlation coefficient (ICC). Each new user should have over 90% agreement as measured by the ICC before proceeding to experimental papers. We explored the fatigue matrix with 10, 20, 30 and 40 images, Users were given the guidance to score these images taking no breaks until they competed the prescribed number.

Statistical Analyses.

All Statistics were done in R using the open-source packages (DescTools, Lme4 and tidyverse). Anova or linear mixed models were used based on data structure to compare groups with the aov function or comparing the fit of linear mixed models using lme from the lmer4 package. Inter reader correlation coefficients were obtained using the ICC function.

RESULTS

Spoti-find captures novel voiding parameters with biological relevance

To expand the analytical utility of the VSA, Spoti-find introduces several new metrics designed to capture spatial and morphological features of voiding behavior. The user interface allows experimenters to adjust specific parameters, select spots using either a rectangle tool that identifies the spots automatically or a polygon tool that allows users to manually outline and identify spots using the fill function (Fig.1A). Users are also able to calibrate volume detection using spots of known sizes (User Manual; see Supplemental Information). Spot binning was implemented to distinguish between primary voids (>22 μL), microvoids (2–22 μL), and nanovoids (1–2 μL) (Fig.1B), based on prior cystometric studies and published estimates of typical voiding volumes(5–8). Spots under 1 μL were classified as noise but remain accessible to users analyzing disease models where smaller volumes may be biologically meaningful (e.g., benign prostatic hyperplasia, BPH (9, 10). Additionally, our tool introduces a circularity metric, reported as a value between 0 and 1, where 1 denotes a perfect circle and 0 reflects a straight line. This metric enables identification of ‘worm-like’ elongated voids—presumed to occur while the animal is in motion—versus rounder, more stationary voids (Fig.1C).

Figure 1.

Spoti-find enables quantification of novel spatial and morphological features in void spot assay analysis. (A) Graphical user interface for Spoti-find is easy to use and adjustable to suit specific experimental parameters. (B) VSA image denoting primary (light purple), micro (orange), and nano voids (pink). (C) Elongated “worm-like” voids and circular voids.

Mice exhibit a strong preference for voiding near the edges of their environment, a behavior thought to be evolutionarily adaptive by helping them avoid open, exposed spaces, a phenotype consistent with their natural thigmotaxic tendencies (5). Increased center voiding due to increased time spent in this sector has been hypothesized as a potential indicator of anxiety-related phenotypes (4). Therefore, we sought to capture the spatial location of each void relative to the paper edge, using the shortest linear distance from the center of each spot to the nearest boundary. This measurement provides a novel proxy for spatial behavioral tendencies in rodent models, leveraging the observation that healthy mice preferentially void along the cage perimeter. Increased central voiding may indicate altered urgency, incontinence, or anxiety-like phenotypes. These novel parameters enable users to extract new behavioral insights from VSA data, without requiring complex computational tools or video tracking systems. By making biologically relevant features like spot shape and location easily quantifiable, Spoti-find enhances the interpretive power of the assay while preserving usability and transparency.

Spoti-find has strong agreement with existing analysis tools and between users

To assess whether Spoti-find performs comparably to existing VSA tools, we evaluated agreement between Spoti-find and the current gold standard, Void Whizzard, for core metrics including total spot count. We used scanned images from a previously performed experiment at the University of Wisconsin in which male C57BL/6J mice at 2 and 27 months of age were measured, some after treatment (manuscript in preparation).

The Void Whizzard analyses were conducted at The University of Wisconsin by researchers with expertise in VSA methodology, using datasets collected under diverse experimental conditions and timeframes. Spoti-find analyses were subsequently performed at The Jackson Laboratory on the same raw images by users without specific training in urologic research. Although Void Whizzard bins voids by diameter and focuses primarily on spot number rather than biologically informed categories, we found strong correlation (R²= 0.9161737 p=2.2x10⁻¹⁶) between tools for spot count between images analyzed in Void Whizzard and Spoti-find (Fig. 2A). The slope of the correlation line (m=0.73) deviated from the predicted m=1, suggesting there is a systematic bias towards a higher spot count with Void Whizzard. We believe this is caused by how Void Wizard and Spotifind handle overlapping spots. Void Wizard is lower throughput requiring manual intervention where more subjective calls are made to parse the overlapping spots. The tradeoff is the ability to process many more images with Spotifind and get more than just count data out, while conclusions from the data are the same as long as all samples are analyzed by the same tool. The strong agreement between Spoti-find and Void Whizzard across these varied datasets highlights the accessibility, ease of use, and analytical reliability of Spoti-find, even when used by non-expert operators. This robustness across institutions, users, and timepoints reinforces Spoti-find’s utility as a widely adoptable platform for VSA analysis.

Figure 2.

Validation of Spoti-find performance against existing tools and across users. (A) Spoti-find and Void Whizzard show high agreement for total number of spots regardless of experimental conditions. Spot sizes/volumes are highly correlated between Spoti-find and Void Whizzard in both naïve animals (B) and treated animals (C), however variation between experimental conditions, imaging settings, lightings, and paper used complicate the ability to compare between different studies. Examples of different paper types, imaging conditions, and spot patterns are shown in (D) and (E).

Correlations between spot volumes in Void Whizzard and Spoti-find were high for both naïve animals of different ages (R² =0.8643384, p=2.495x10⁻¹⁵) (Fig. 2B) and animals of the same age under different treatment conditions (R²=0.7805848, p=0.007707.) (Fig. 2C). However, despite strong agreement in spot counts, correlations in volume were less robust when comparing across experimental conditions (R²= 0.6233786, p=1.726x10⁻⁷). This discrepancy likely stems from variability in imaging factors such as lighting, camera settings, and paper types, highlighting the need for caution when comparing historical datasets collected under heterogeneous conditions (Fig. 2D–E). For example, in Fig. 2B an approximately 40 uL void corresponds to ~4000 arbitrary volume units (AVUs), whereas under different imaging conditions (Fig. 2C) a void of similar size corresponds to only ~300 AVUs. To account for such inter-experimental variability, all Spoti-find volume measurements are reported in AVUs, since datasets might be acquired across multiple laboratories using diverse imaging hardware, paper types, and acquisition parameters. These differences preclude the use of a universal calibration curve to convert pixel-based measures into absolute microliter values. Instead, Spoti-find estimates volume as a function of spot area and pixel intensity, both derived from pixel data and thus highly sensitive to acquisition settings. While spot counts are robust to these variables (Fig. 2A), volume comparisons across datasets lacking consistent imaging conditions should be interpreted in relative rather than absolute terms (Fig. 2B–E).

To determine the practical robustness of Spoti-find for large-scale dataset analysis, we tested whether user fatigue affected performance across prolonged analysis sessions. For this, we used images from both male and female C57BL/6J mice at 18 months of age performed at The Jackson Laboratory.

Individual users were asked to analyze batches of 10, 20, 30, or 40 papers without taking breaks. Across multiple trials, we found that users were consistently able to complete up to 30 papers per session before experiencing accuracy declines or noticeable fatigue, whereas 40 papers proved unsustainable. We note that this threshold may be model dependent; for instance, diabetic mouse models tend to produce more frequent, smaller voids that require more manual input than models in which voiding is dominated by fewer large-volume primary voids.

We also evaluated inter-user agreement to assess how user variability may impact Spoti-find results. Most parameters showed high concordance across users (Supplementary Table 1), demonstrating that the tool is generally robust to individual analytical styles. However, variability was higher in parameters like circularity and average nano/micro void volume, which are more sensitive to minor differences in polygon selection. For example, small linear segments introduced while tracing void perimeters can disproportionately affect circularity in small spots due to their lower overall circumference. These findings underscore the importance of consistent technique and cautious interpretation when drawing conclusions from highly granular parameters.

Novel and known age-related alterations in voiding behavior and clinically relevant aging features are captured by Spoti-find analysis

Urinary dysfunction (including increased frequency, reduced bladder capacity, and high post-void residual volume) is a well-documented consequence of aging in both humans and rodent models and behavioral assays like the VSA have proven sensitive to these changes in strain- and age-dependent studies (1, 5, 7, 11). Additionally, bladder diaries, which record frequency, volume, and voiding patterns over several days (11), remain a first-line clinical tool for assessing urinary dysfunction in older adults. They routinely capture hallmark age-related changes such as increased frequency of small-volume voids, larger but less frequent primary voids, and episodes of urgency or incontinence. The VSA serves as a direct preclinical counterpart to this assessment, and Spoti-find allows us to extract features that closely align with diary-based readouts.

Aged C57BL/6J mice exhibited greater variability in spot counts across all size bins, with the most pronounced differences observed in the nano range (primary p=0.0016, micro p=0.0105, nano p<0.001) (Fig.3A). These patterns reflect the emergence of small-volume spotting events that are characteristic of aged models. Unsurprisingly, aged animals displayed higher primary void volumes relative to young controls, consistent with their larger body size and bladder capacity at this age, while micro and nano void volumes were comparable between groups (primary p=0.0099, micro p=0.4108, nano p=0.7430) (Fig. 3B), likely due to the narrow absolute range of these bins. Circularity analysis provided a proxy for voiding intention: primary voids were more circular in aged animals, no differences were observed in the micro bin, and nano voids were less circular in aged animals (primary p=0.0843, micro p=0.8944, nano p=0.1105, total p<0.0001) (Fig. 3C). This suggests that while primary voiding behavior remains intact with age, small-volume voids are less intentional in nature. When averaged across all size bins, aged animals exhibited an overall reduction in circularity (Fig. 3C), supporting the interpretation that Spoti-find can detect incontinence-like phenotypes previously described in these models. Notably, this pattern is consistent with cystometric findings showing that aged animals can perform complete primary voids (5, 6) but also exhibit additional low-volume leak events, sometimes reported as a “dysfunctional responder” phenotype. Finally, the average distance to the paper edge was greater in aged animals compared to young controls (p<0.0001) (Fig. 3D), further highlighting the maladaptive voiding behaviors that arise with age. Collectively, these data demonstrate that Spoti-find captures both expected and novel age-associated changes in voiding behavior across multiple analytical dimensions.

Figure 3.

Spoti-find detects voiding behavior differences attributable to aging. Comparison of void counts by spot size (A) and volume by spot size (B) between young and old mice. (C) Average total circularity and circularity by spot size between young and old mice. (D) Average spot distance to paper edge between young and old mice.

An ongoing debate in the field is whether to acclimate the animals by repeating the VSA over several days and discard the measurements of the first (few) days. We explored this using a group of C57BL/6J males and females and performing VSA on all animals on four consecutive days. Our data on total spot counts shows there are some small statistically significant differences in the females between Day 1 and Days 3 and 4 (Fig. 4). These differences are much larger in the males where, in addition to the number of spots, the variability between animals decreases over the four days. For spot volume, we see a similar picture for the females but no difference in the males (Fig. 4). Overall, we recommend acclimating the animals and only use Day 3 and Day 4 measurements with VSA.

Figure 4.

Measurements over four consecutive days shows changes in spot count and spot volume, suggesting an acclimation effect

DISCUSSION

Spoti-find represents a next step in the evolution of VSA analysis, building on the success of Void Whizzard, which brought structure, reproducibility, and accessibility to the field. By introducing new biologically relevant parameters such as circularity, void size binning, and distance from the paper edge, Spoti-find expands the scope of insight that can be extracted from this inexpensive and non-invasive assay. While traditional metrics like total spot count and voided volume remain valuable, they often overlook the behavioral (18) and spatial (1) nuances of voiding that may be particularly relevant in models of aging, neurodegeneration, or systemic disease. Spoti-find offers a path toward integrating these dimensions, enabling deeper phenotyping and improved translatability.

Importantly, we see Spoti-find not just as a new tool, but to make more of the data we already collect. The VSA is uniquely suited to longitudinal, non-terminal assessment of urinary behavior in rodents. It can be performed repeatedly, even across the full lifespan of an animal, with minimal handling and without the need for surgical instrumentation like assessments of urinary physiology directly. This makes it especially powerful in models where animals are scarce or valuable, such as neurodegenerative disease models or lifespan cohorts, and aligns with the broader scientific goal of reducing animal numbers while increasing data yield per subject. Because Spoti-find makes it easier to quantify more parameters from each assay, it extends the utility of existing datasets and improves the granularity of behavioral phenotyping without increasing experimental burden.

We also highlight several critical experimental insights enabled by Spoti-find. Acclimation time was found to be of particular importance, as mice, regardless of sex, require at least two full days in their housing cages before voiding behavior stabilizes. This finding underscores the need to account for behavioral adaptation when designing VSA studies and suggests that early time points may not reflect true baseline phenotypes. To maximize the biological value of VSAs, it is critical not only to refine how data are collected, but also how they are analyzed. Spoti-find supports output formats compatible with more rigorous statistical approaches, including linear mixed models (LMMs), which allow for greater sensitivity and flexibility compared to traditional summary statistics. The typical method of averaging across replicates and then averaging across animals (commonly referred to as the “average of averages” approach) risks obscuring biologically relevant variability and inflating false significance due to inappropriate aggregation (19). In contrast, LMMs retain within-subject variance across repeated measures, account for random effects (such as cage or user), and more accurately reflect the hierarchical structure of preclinical behavioral data. These advantages are particularly relevant for urinary phenotyping, where spot number and distribution can vary meaningfully across time and individuals. This statistical improvement, when paired with Spoti-find’s detailed output structure, allows researchers to make more rigorous, nuanced inferences from their data, an essential step for behavioral studies with subtle or distributed phenotypes.

It is important to note that Spoti-find cannot automatically quantify overlapping void areas (‘spots within spots’). This limitation may affect the interpretation of circularity measurements, as overlapping voids may appear less circular than they are. Users should be aware that highly eccentric or irregular void shapes may represent either genuine irregular voiding patterns or overlapping voids that could not be fully resolved. Manual polygon selection can help address some of these cases, but complete separation of all overlapping voids is not always feasible. This represents a known limitation of the current analysis approach and should be considered when interpreting void morphology data. This remains a high-priority area for future development and incorporating this feature would increase the accuracy of voided volume and bin classification, particularly in models with high voiding frequency. Full automation of VSA analysis remains a difficult challenge due to variability in paper type, lighting conditions, spot contrast, and image quality between datasets and experiments performed at different institutions. For these reasons, Spoti-find was built to balance automation with user control. The polygon tool allows for precise manual correction where automation falls short, and adjustable thresholds support flexible application across experimental contexts.

One of the most exciting aspects of Spoti-find is its ability to approximate behavioral features that previously required costly or low-throughput methods like metabolic cages or video tracking. Intentional voiding behavior—such as stopping, squatting, and releasing a coordinated void—is difficult to detect without invasive or intensive monitoring setups (20). By introducing metrics like circularity and spatial distribution, Spoti-find allows researchers to infer aspects of intentionality and urgency using standard cage conditions and a single sheet of paper. This dramatically increases throughput, preserves the natural environment, and makes behavioral voiding analysis more accessible to a wider range of labs and study designs.

The development of Spoti-find was made possible through generous, cross-institutional collaborations that shared pre-existing datasets, which enabled optimization of the tool, reproducibility testing, and blinded benchmarking comparisons against Void Whizzard. Spoti-find is being released under an open-source license that encourages modification and community-driven improvements, and this tool is intended as a flexible platform that researchers can adapt, refine, and build upon in the spirit of open science. Given the prevalence of urinary dysfunction across diseases of aging and the nervous system (21), Spoti-find will be broadly useful for identifying and characterizing phenotypes that are often underappreciated in mouse models. Efforts to improve the clinical relevance of behavioral assays -- particularly in areas of urinary function and behavioral control, where cross-species anatomical and neurological differences can obscure interpretation -- are essential for translating preclinical findings into human benefit. By equipping researchers with new ways to analyze, standardize, and quantify VSA data, Spoti-find aims to advance both urologic and broader biomedical research. In the clinic, bladder diaries are used to track symptoms such as frequency, urgency, leakage, and incontinence (22) -- many of which are behaviorally complex and not captured by void count or volume alone. Incorporating features like circularity, which may reflect intentional versus passive voiding, and distance from the edge of the cage liner, which may relate to voiding urgency or anxiety-like behavior, Spoti-find allows researchers to approximate aspects of urinary dysfunction that are routinely assessed in human patients, thereby increasing the biological resolution and clinical relevance of the VSA. While the manual interface limits throughput and may introduce variability (particularly in small or finely structured voids), these limitations are offset by the depth of exploratory phenotyping it enables, complementing high-throughput solutions like ML-UrineQuant (3). As Spoti-find is applied to additional neurodegenerative disease models (such as amyotrophic lateral sclerosis (ALS) (23), Parkinson’s disease (24), and Alzheimer’s disease (25), all of which are associated with well-documented urinary phenotypes in humans), it is poised to yield insight into dynamic brain-bladder axis interactions through longitudinal, non-invasive assessments that preserve downstream experimental endpoints while informing mechanistic follow-up using standard physiological assessments (such as pharmacomyography and cystometry) to fully understand the changes responsible for altered voiding behavior.

Supplementary Material

Supplemental Table S1. 10.6084/m9.figshare.31707163

DATA AVAILABILITY

The datasets used in this study have been deposited at Images.jax.org specific links to each dataset and a brief description of each is below.

Validation Dataset: A dataset of 28 Images used to assess inter-reader correlation coefficients and supplied with scores from three people for users to benchmark their spot identification. Images available at: https://images.jax.org/webclient/?show=dataset-5501.

10-Image Dataset: A set of ten images from males and females on a C57BL/6J background. Images available at: https://images.jax.org/webclient/?show=dataset-5506.

20-Image Dataset: A set of twenty images from males and females on a C57BL/6J background. Images available at: https://images.jax.org/webclient/?show=dataset-5502.

30-Image Dataset: A set of thirty images from males and females on a C57BL/6J background. Three readers scored this dataset without breaks. Images available at: https://images.jax.org/webclient/?show=dataset-5503.

Aging Dataset: A set of images from 25 young (2 months) and 25 old (27 months), male C57BL/6J mice from the Ricke Lab. Images available at: https://images.jax.org/webclient/?show=dataset-5505.

CODE AVAILABILITY

Code for Spoti-find is available at https://doi.org/10.5281/zenodo.18622284.